USA 2016 presidential election – with better voting methods, the results would have been different

By Warren D. Smith, 3 Dec. 2016

Table of Contents

  1. Official results
  2. Popular versus Electoral vote
  3. Was Trump's victory fraudulent and the result of "hacked" voting machines/computers in "swing states"?
  4. The "debates" and other rigged games
  5. Were Gary Johnson and/or Jill Stein "spoilers"? No.
  6. Was Clinton a spoiler? Yes.
  7. Instant Runoff Voting (IRV) would not have altered the winner
  8. This election was a huge distortion of democracy
  9. Enormous costs incurred by USA's foolish use of plurality voting system
  10. Was Trump's victory a "huge surprise"? Not to me and some others
  11. And now for something completely different: Lichtman's "13 keys" to the white house
  12. Approval voting (with "primary" stages eliminated by merger so just one combined election)
  13. Score voting (with "primary" stages eliminated by merger so just one combined election)
  14. Approval voting (official post-primary contenders only)
  15. Score voting (official post-primary contenders only, but sometimes a few wildcards added)
  16. Suicide-level idiocy of both the Republican and Democratic (and Green!) parties
  17. **Conclusions**

Official results

The USA (8 Nov. 2016) presidential election was won by Donald J. Trump (Republican party). This table shows the preliminary (as of 3 Dec; all figures are expected to change slightly as further vote-counting occurs, and the electoral college will meet on 19 Dec.) official election results.

Candidate (Party)Popular voteElectoral vote
Donald J. Trump (Repub)62916K (46.0%)306*
Hillary R. Clinton (Demcrtc)65750K (48.1%)232
Gary Johnson (Librtrn)4460K (3.3%)0
Jill Stein (Green)1440K (1.1%)0
Others (incl. write-ins, all combined)2063K (1.5%)0
TOTAL136 million (100%)538

*Update: Actually due to "faithless electors" who voted for Bernie Sanders, John Kasich, Colin Powell, Ron Paul, and Faith Spotted Eagle, neither Trump nor Clinton got the exact numbers of electoral votes tabulated above (which they would have gotten with faithful electors), getting 304 and 227 respectively.

This actually was the largest number of people winning positive numbers of electoral votes seen in 200 years, and the greatest number of "faithless electors" since the US civil war.

270 electoral votes sufficed to win. The eight closest states (by Trump-Clinton difference as percentage of that state's total vote count) were

State (Abbrev)EVsWinnerTrump-Clinton marginJohnsonStein
Michigan (MI)16Trump0.23%3.59% 1.07%
New Hampshire (NH)4Clinton0.36%4.12%0.87%
Pennsylvania (PA)20Trump 0.73%2.40% 0.82%
Wisconsin (WI)10Trump0.81%5.00% 1.13%
Florida (FL)29Trump1.20%2.20% 0.68%
Minnesota (MN)10Clinton1.52%3.84% 1.26%
Nevada (NV)6Clinton2.42%3.32% (0)
Maine (ME)2Clinton2.68%5.10% 1.90%

The above table also shows, in the rightmost columns, the percentages of voters in those states who chose the Libertarian Party's candidate Gary Johnson and the Green Party's Jill Stein. (Stein was not on ballot in Nevada.)

Popular versus Electoral vote

In most other countries, there is only (what the USA calls) the "popular vote" – they just call it the "vote." Under those rules Hillary Clinton would have won by about 2.9 million votes. But in the USA, the candidates win individual states, and then those states contribute various integer numbers of electoral votes, then whoever gets a majority of those wins. Hence Trump won. If nobody has a majority, the House of Representatives decides the election.

If fewer than 50K votes had switched Trump→Clinton in the crucial states, then she would have won, even if meanwhile tens of millions of votes were switched Clinton→Trump in irrelevant states.

Somewhat over 90% of the time, historically, the popular and electoral winners have agreed, but not in 2016. The other candidates who won the USA popular vote but lost the election thanks to the "electoral college" and/or House were: Andrew Jackson in 1824, Samuel Tilden in 1876, Grover Cleveland in 1888, and Al Gore in 2000.

It is worth briefly revisiting that historical roll to point out a remarkable thing: Donald Trump's was by far the greatest ever popular vote loss by a presidential winner. The table explains what we mean by that.

YearCandidate A Candidate B A-B pop. vote margin B-A electoral vote margin
2016H.R.ClintonD.J.Trump 2.9 million=2.1%306-232=74=13.75%
1876S.J.TildenR.B.Hayes 252666=1.0%185-184=1=0.27%
2000Al GoreG.W.Bush 543816=0.5%271-266=5=0.93%
1888Grover ClevelandBenj.Harrison 94530=0.9%233-168=65=16.2%

In this table, "A" is the popular vote winner, while "B" is the electoral vote winner (hence elected president). The only election that approaches challenging the supremacy of Trump-Clinton 2016, is Hayes-Tilden 1876 which was fraudulent! The story of how that came about is described in many history books, but let us merely quote

There is no longer any doubt that this election [1876] was "stolen." – Samuel E. Morison, The Oxford History of the American People (Oxford Univ. Press 1965) p.734.

There is also one more election, 1824, that belongs on this list as a special case. Andrew Jackson won both the popular vote (to the partial extent that the popular vote was even counted at that time; 6 states did not count it) and the electoral vote, but still lost the Presidency to J.Q.Adams due to the intercession of the House of Representatives. This was the only time in US history that nobody got an electoral vote majority, unless you count 1800 where Jefferson & Burr were initially tied. (Jackson could have clinched a win by getting 131 electoral votes, but only got 99 hence this election was decided by the House. Clay, who was Speaker of the House, threw his support to Adams. Crawford had suffered a paralyzing stroke that put him out of contention, and to the extent he had an effect, he tended to split the House vote with Jackson.)

     Candidate            Pop.Vote    Electoral Vote
    Andrew Jackson         151271         99
    John Quincy Adams      113122         84
    William H. Crawford     40856         41
    Henry Clay              47531         37

Both the 1876 and 1824 elections were regarded far and wide as travesties and had very thunderous and damaging historical repercussions. In particular, the price the Republicans agreed to pay to get the Democrats to accede to the corrupt inter-party bargain solidifying Hayes' fraudulent 1876 election was basically to cease enforcing legal rights for blacks and to open the floodgates to "Jim Crow" oppression over the next 90 or so years.

Jackson regarded Clay↔Adams as having made a "corrupt bargain" to undemocratically deny him the presidency and also blamed the death of his wife on slanders by his political opponents. As a result when Jackson did become President 1829-1837 he had the attitude that anything he did to gain power, regardless of the rest of the government, was OK. He was a total slash-and-burn maniac, starting the so-called "spoils system" whose corrupt tendencies took over the US government, to a too-large degree, from then on; and presiding over the Cherokee "trail of tears" genocide in defiance of a Supreme Court ruling.

Trump's electoral college margin in 2016 ranked 46th-from-top (percentagewise) out of the 58 USA presidential elections according to data compiled by Prof. John Pitney, while his popular vote loss was by far the largest among all Presidents. Trump described this as

I mean, think of it. We won in a landslide. That was a landslide. And we didn't have the press. The press was brutal. You know what? Hey.
  – Donald J. Trump speaking in Cincinnati Ohio, 1 Dec. 2016, at the first of 10 planned post-election "victory rallies."

Was Trump's victory fraudulent and the result of "hacked" voting machines/computers in "swing states"?

As of the date of this writing (Nov. 2016) I certainly do not know. Some prominent computer security and voting experts – e.g. J.Alex Halderman, Philip Stark, Ron Rivest, and Barbara Simons – claimed, due to circumstantial evidence, that this was a likely-enough possibility that "forensic audits" ought to be conducted in those states; and urged the Clinton team to demand them. Their evidence mainly consisted of

  1. A claim that the parts of Wisconsin using "computerized voting machines" gave Trump about 7% more votes.
    See also: this report by Richard Hayes Phillips pointing out more suspicious facts about Wisconsin's numbers, for example he lists 7 towns in WI with turnouts exceeding 100%... A 59% majority of the vote-counting machines in Michigan's most-Democratic area, namely 87 optical scanners, broke on election day [Chad Livengood & Joel Kurth: Half of Detroit votes may be ineligible for recount, The Detroit News 6 Dec. 2016], causing 392 of the 662 Detroit precincts to announce (what the state later admitted were) incorrect ballot totals, i.e. with the number of voters disagreeing with the number of votes. Daniel Baxter, the elections director for the city of Detroit, blamed aging equipment and summed up the situation with "It's not good." ... Michigan then claimed it was illegal for those 392 precincts to be recounted!
  2. The fact many US intelligence agencies had announced they believed Russia had tried to influence the election by cyber methods, including breaking into both the RNC and DNC computers and email systems, and (separately) Hillary Clinton's campaign manager John Podesta's emails, which they then leaked to the public. They also broke into voter registration databases in at least two states (Illinois and Arizona) to steal voter data, and tried to do so in at least 20 states. This was according to several US intelligence sources speaking to the press anonymously, as well as FBI director James Comey. A large number of US intelligence agencies came to the "high confidence" conclusion the actor behind this was Russia. Then the CIA (and according to leaks to the Washington Post in mid-December, also later the FBI) further concluded this was directed by Putin and/or the "highest levels of the Russian government" and with the goal of electing Trump. Reuters (15 Dec 2016) then disclosed U.S. election agency breached by hackers after November vote and the hacker was Russian-speaking. Incidentally, US intelligence as of 2016 believed Putin was the world's richest man ($85 billion) as a result of massive "kleptocracy" of Russian state assets. (His official salary, in contrast, was below $200K.) There also seemed little doubt Putin had had many of his opponents murdered.
  3. The fact the pre-election polls seemed to be systematically more pro-Clinton and anti-Trump than the official vote totals.

Even Halderman said that he thought this evidence was weak – e.g. he felt a priori it was more likely there had been no hacking – but these experts felt that the net impact of this evidence was great enough to justify auditing. As Halderman put it: "I don't believe that either one of these seemingly unlikely explanations [hacked election or just an innocent Trump-favoring fluctuation] is overwhelmingly more likely than the other."

But despite whatever pleas these experts pled, Clinton refused to act.

Jill Stein, the Green party candidate, then announced that she was going to pay for recounts in Wisconsin, Michigan, and Pennsylvania, (and perhaps Ohio?), and raised over $4 million in 1 day to pay for it! Her total fundraising effort was planned to be about $8 million. (That 1-day fundraise was actually more money than she earlier had raised to fund her entire presidential campaign.)

Donald Trump had some peculiar reactions to this. First, he called Stein's recount effort a "scam" and "ridiculous," e.g. "This is a scam by the Green Party for an election that has already been conceded, and the results of this election should be respected instead of being challenged and abused" (25 Nov). He indeed then filed at least one objection and one lawsuit seeking to block it; and then Federal and Michigan State courts simultaneously announced conflicting rulings that Michigan should and should not recount (causing both sides to declare victory, and total confusion). Second, on 27 Nov, Trump claimed that not only did he win "the Electoral College in a landslide, I won the popular vote if you deduct the millions of people who voted illegally." He shortly thereafter added: "Serious voter fraud in Virginia, New Hampshire and California – so why isn't the media reporting on this? Serious bias – big problem!"

My reactions to Trump's claims are mixed. On the one hand, I feel we should take him seriously since he is the president-elect and has access to far more vote data and expertise than I. Indeed he by now must be one of the top few experts in the USA. On the other hand, Trump provided zero evidence for these claims, and note the documented fact that 70% of whatever Trump says that journalists check turns out to be "mostly false" or worse, with over half being 4-Pinocchio lies.

And, oddly enough, as of 1 Dec. 2016, there have been exactly 4 cases of voters casting fraudulent votes in the 2016 election reported in the press – and all four were registered Republicans:

  1. Terri Lynn Rote first early-voted Trump, then voted again at a different polling site in Des Moines Iowa. She was arrested on 27 Oct. This was "the first time in 12 years that Polk County Auditor Jamie Fitzgerald can remember ever having to report potential voter fraud." [Charly Haley: Voter fraud suspect arrested in Des Moines, Des Moines Register 29 Oct 2016]. Rote told Iowa Public Radio that she did it because she was afraid her first ballot for Trump would be changed to Hillary Clinton, because "the polls are rigged." She could be imprisoned for up to 5 years.
  2. Phillip Cook was arrested on Election Day after voting twice in Sugar Land, Texas. He claimed to be an employee of Trump's campaign who was "testing the security of the electoral system." But the Trump campaign denied that.
  3. Audrey Cook was a Republican election judge in Alton Illinois, who was charged on 4 Nov. with double voting. She told the Associated Press she'd filled out absentee ballots for both herself and her dead husband Vic because she knew he would want Donald Trump to be president. [Alton election judge charged for allegedly casting ballot for dead husband, Associated Press, 4 Nov 2016].
  4. Gladys Coego was hired to open absentee ballots in Miami-Dade County. One of her co-workers noticed that she was going a step further, filling in the bubble for mayoral candidate Raquel Regalado (R) with a pen she had in her purse. She was caught in the act and arrested. [Patricia Mazzei: Two women busted for election fraud in Miami-Dade, Miami Herald 28 Oct. 2016].

In any event, as of the present writing, neither Stein nor Clinton has made any claim that the election was fraudulent or hacked. But Trump did. Indeed, for Trump's claim to be correct, there must have been at least about 4.5 million illegal voters, if they voted 80:20 for Clinton over Trump. If all were caught and imprisoned, that would more than double the USA's prison population, and exceed the number of illegal-voting convictions in all previous US history, combined, by a factor of order 1000.

So to summarize: (i) Stein has called for a recount, (ii) prominent computer scientists and fraud experts have called for a forensic audit, (iii) Clinton has agreed to provide support to help Stein's effort, and (iv) Trump has denounced the official election results as enormously fraudulent. So there would seem to be unanimous agreement recounts and forensic audits are somewhere between "good ideas" and "absolutely necessary." Unfortunately I have doubts Stein will be able to cause a genuine forensic audit, as opposed to a mere garden variety recount.

Update mid-December 2016: Stein succeeded in causing a recount in Wisconsin, which changed its totals by less than 1000 votes (e.g. its Trump-Clinton margin widened by 162) and did not appear to have any "forensic" component. Her attempts to recount Pennsylvania or to have experts examine its voting machines for "hacks" both were blocked. In Michigan, the closest state, Trump supposedly beat Clinton by 10704 votes out of 4.8 million cast. A recount began but was aborted by court order. But 75335 votes were cast but never counted by Michigan's optical-scan machines. Many of these votes could have been counted by humans, e.g. because they had been filled out in red ink unreadable by the machines, or using "X" marks or ☑ symbols instead of filled-in-circles. And Stein provided the money to do it. But they remained uncounted by order of courts and/or Michigan Attorney General Bill Schuette. Most of these uncounted votes occurred in Detroit and Flint, two Clinton strongholds. Also note, any precinct where the number of ballots differed from the number of voters (e.g. 392 of the 662 Detroit precincts!) was forbidden to recount, and any precinct where seals on voting machines were broken, ditto. This is the exact opposite of what any "forensic audit" would do, and a dream come true for any fraudster.

In any case, for the purposes of the rest of this page, we are going to take the attitude that the election results were legitimate. (And if the result really were fraudulent due to fairly small alterations in "swing states," that would not affect the validity of almost all we shall say.)

The "Debates" and other rigged games

There were three highly publicized and televised debates between Hillary Clinton and Donald Trump, organized by the "Commission on Presidential Debates"; transcripts of them are available on the CoPD website.

There also was a considerably less publicized 3-part debate between Jill Stein and Gary Johnson, organized by Tavis Smiley. Its first two parts were televised by PBS and all three were made available over the internet I, II, III.

Date LocationModeratorWinner (Polls, but online "polls" allowing anybody to vote as many times as they want, not included. You are warned that "debate-watchers" are a self-selected, and not a random-uniformly selected, subset of "all people.")
26 September Washington University, St. Louis, Missouri. Anderson Cooper. 84M viewers. Clinton by 62-27 in CNN/ORC telephone poll of 521 debate watchers conducted immediately after; and by 51-40 in a PPP poll of 1002 pre-agreeing watchers; and by 57-30 in a YouGov survey questioning 1154 US adults who watched it.
9 October Hofstra University, Hempstead, New York. Lester Holt. 66M viewers. Clinton by 57-34 in CNN/ORC telephone poll. Also by 47-22 in a YouGov poll interviewing 812 registered voters who watched debate; and 42-28 in a Politico/MorningConsult poll released 11 Oct.
19 October University of Nevada, Las Vegas Nevada. Chris Wallace. 72M viewers. Clinton by 52-39 in CNN/ORC telephone poll of 547 debate watchers. Also 49-39 in a YouGov poll interviewing 1503 registered voters who watched debate, and the exact same 49-39 result was found in an independent CBS battleground-state poll interviewing 943 over internet.
1 November Los Angeles, California. Tavis Smiley. (1M viewers?) No poll done to assess any "winner." But in my subjective opinion, Johnson performed better than any of the three performances by Trump. Watch and judge for yourself.

The CoPD is an inherently corrupt body since it is a corporation created jointly by the Democratic and Republican Parties, not an independent body owned by the media. These two parties, not the media or a public interest group such as the League of Women Voters, make the rules, and in particular those rules always make sure that no third-party candidate is allowed to debate, unless both happen to want him to that year. If any third-party candidate shows up at the debate site, even with a legally purchased ticket, the Republican and Democratic parties have so far, every time, gotten armed thugs (oh sorry, I meant police) to escort him/her offsite to block his/her attendance.

For example, after that happened to presidential candidate Ralph Nader in 2000, he sued, and the suit was settled in April 2002 with the CoPD paying him an "undisclosed amount of money" and issuing a hilarious letter of apology to him. It explained that the CoPD had misunderstood Nader's intentions and if they'd realized he merely intended to use his ticket to attend (the way he said he would to the arresting officers) they would have been happy to help in any way they could! Why was it that no other ticket-holding attendee of the debate was arrested, and no other had their intentions misunderstood? The letter did not say.

Did that experience cause the CoPD to cease these practices? No. In 2012, Jill Stein, the Green Party presidential candidate, and her vice-presidential candidate Cheri Honkala, were arrested by police as they attempted to enter the grounds of the presidential debate site at Hofstra University, then both were detained for eight hours handcuffed to chairs in a warehouse, guarded by 13 police, with no phone call allowed, then finally spirited away in a Secret Service car. In 2016 essentially the same thing happened. (Oddly enough, no charges were filed against Stein or Nader. They were just arrested, not charged.)

Note that Clinton won all three debates by large margins, although Trump improved relative to her each time. (By extrapolating the improvements, presumably Trump would have first won a hypothetical 5th debate?)

Trump described these as "I won every poll [about the first] Presidential Debate – except for the little watched @CNN poll" [tweeted 27 Sep.] and he "Won every poll. Virtually every poll." [told to his supporters later that same day]; and he also won "the second debate in a landslide (every poll)" [tweeted 11 Oct.] and finally Trump told Bill O'Reilly on Fox News (Oct.27) that he had "won the third debate easily" and all the online polls had him winning, but then "dirty polls" came out showing him "losing by numbers that were ridiculous" which was since pollsters are engaged in "suppression" and "tremendous dishonesty."

In addition to the debates we just tabulated, there also were 12 debates between the Republican primary contenders only, and 9 between the Democrats only.

The 9 Democratic debates were watched by an estimated 4.5 to 15.8 million viewers each, averaging 8M. The 12 Republican debates had 11 to 24 million viewers each, averaging 15.5M. Impressively, even the least-watched Republican debate got more viewers than 8 among the 9 Democratic debates! This was despite the facts that in 2016 more Amercians identified as Democrats than Republicans (Dem=35%, Rep=28% according to Pew; Dem=29%, Rep=26% according to Gallup) and the Democrats had fewer debates, both of which would naively be expected to get them more viewers.

Why this viewership discrepancy? The Democratic debate schedule appeared to have been chosen either by somebody tremendously more idiotic than whoever scheduled the Republican debates – or actually with the intent of causing low viewership. It was widely speculated that the latter was the case. Why? The hypothesis was that the DNC was not serving as an unbiased referee, but rather was biased pro-Clinton. And it was believed a priori that Clinton would begin the primary in the lead and with the most money, and that the more debating she did, and the more visible it was, then the more chances her rivals would have. (E.g. with no debates at all, or no viewership, she'd have the best chances to retain her initial lead.) Therefore, the Democratic debate schedule was intentionally set up ahead of time to help Clinton.

That conspiracy theory was fueled by several other facts, all seeming compatible with it. Wikileaked emails from within the DNC unveiled the facts that

  1. DNC communications chief Luis Miranda, instructed his staff to put out, in a "unattributable manner," allegations that Bernie Sanders' supporters were engaged in acts of violence.
  2. There also were email discussions within the DNC about exposing – or, rather, claiming – that Bernie Sanders was an atheist, because that hopefully would hurt his popularity in the South.
  3. Donna Brazile, then a journalist with CNN, emailed Clinton Communications Director Jennifer Palmieri on 12 March that "From time to time, I get the questions in advance." and then used the rest of the email to inform Hillary Clinton ahead of time of the wording of a question about the death penalty to be asked her at a CNN-run "town hall" (although ostensibly such questions were unknown to the candidates ahead of time). The question then indeed was asked at the town hall on 13 March, with quite similar wording, including re-using the numbers "1414," "1973," "156," and "1976"; the words were not exactly the same, but it later was found that the question had undergone editing and originally had been exactly the same as in Brazile's email. Brazile was fired by CNN on 31 Oct. after this story broke, but was immediately hired by the DNC as "interim chair."
  4. The head of the DNC, congresswoman Debbie Wasserman-Schultz, resigned during the Democratic National Convention in July after this email/bias scandal broke. However, she was immediately hired by Clinton's presidential campaign.
  5. Wasserman-Schultz shut down Sanders' access to DNC data and had to be sued before restoring it; also she conveniently overlooked (despite a complaint filed by Sanders following a Washington Post story about it) what seem to me to be blatantly illegal use by Clinton of the DNC as a money-launderer to permit her to effectively evade campaign-finance donation limits.
  6. Both the Sanders and O'Malley campaigns accused the DNC of acting in a pro-Clinton biased manner, rather than their ostensibly unbiased role, during the primaries. E.g. Lis Smith, a former aide to primary contender Martin O'Malley, told The Atlantic. "The emails just confirmed what we already knew; [Wasserman-Schultz] was not an honest broker in this process. She should have been gone long ago." And Sanders told NBC, "Nobody has apologized. But this [scandal] does not come as a surprise to me or my supporters. There is no doubt that the DNC was on Secretary Clinton's campaign from day one."

I do not know whether the pro-Clinton bias of the DNC was enormous or small, but it is clearly proven that, in at least some cases, it existed.

The DNC emails were widely claimed (without any published proof; the evidence was secret) to have been leaked by "Russian hackers" ultimately directed by Vladimir Putin. Wikileaks and its head Julian Assange have never revealed any of their sources and indeed often do not know their sources. However, 27-year-old DNC staffer Seth Rich was murdered on 10 July (shot in the back while walking on a street in Washington DC; he was not robbed and the motive and killer remain unknown). Assange then implied in a television interview that possibly Rich might have been the source or somehow involved, and in any case Wikileaks on 9 August offered a $20000 reward for information leading to the arrest of Rich's killer.

As of this writing, it has not been collected.

Wikileaks head Julian Assange, and another Wikileaks figure Craig Murray, both claimed that their source was not either Russia nor indeed any "state actor" and claimed that they both knew (in this particular case) the identity of the leaker.

To explore ways in which the DNC and "Democratic establishment" may have put their thumbs on the scales to favor Hillary Clinton over her top rival (for the Democratic nomination) Bernie Sanders, let us examine the two biggest Democratic-party-controlled states: California and New York.

A policy of the DNC and the completely-Democrat-controlled California government, which definitely hurt Sanders tremendously versus Clinton, was uncovered by investigative journalist Greg Palast. The issue is as follows. Sanders had a big advantage among "independent" voters, while Clinton was preferred by "registered Democrats." The former were more numerous; and not-coincidentally, USA-wide, Sanders was preferred over Clinton. However, what mattered for the purpose of winning the Democratic party's presidential nomination was not the USA-wide preference, but rather the preferences of the small subset of the USA which voted in the Democratic primaries. In "open primary" states, anybody is allowed to vote in any party's primary, for example a Republican, Green, or Independent could vote in the Democratic primary (but you must vote in at most one party's primary). In "closed primary" states, only registered Democrats could. And in California, registered independents ("NPPs") were allowed to vote in the Democratic (but not Republican!) party primary.

California voter registration statistics as of 24 Oct. 2016,
           and poll claims about their Clinton:Sanders preferences
Party Type            Number |                    LAtimes   Field
Registered Democrat    8720K |  Clinton:Sanders by 44:43%  49:40%
Registered Republican  5048K |  (Sanders>Clinton?)
Registered NPP         4711K |                     34:50%  27:54%
Other                   932K |  (Sanders>Clinton?)
Totals                                 Sanders by 1, Clinton by 2

The claimed Clinton-Sanders preferences are from a USC Dornsife/Los Angeles Times poll surveying 1500 voters (19-31 May 2016) which included 903 Democratic and 503 "likely Democratic" voters; and the Field Research Corp. poll 26-31 May consisting of 1002 (79% Democrat and 21% NPP) likely voters. There also was a Marist/WSJ poll (29-31 May) which found either a 2 point lead for Clinton or a 1 point lead for Sanders, depending on how it is interpreted. In short, all three polls agreed the race was very close; but obviously, by far the most crucial question was what the NPP/Dem composition of the Califormia Dem-primary electorate, was going to be.

The answer was revealed by the 7 June official vote total reported by the LA Times – a huge 55.8:43.2 for Clinton, vastly outside the claimed margins of error of any of these polls! [UPDATE: 5 weeks later, claimed a revised count found Clinton had only won by 53.1 to 46.0 percent.] So evidently, the NPPs were heavily excluded from the voter pool, to a far greater degree than any pollster expected.

Why? Well, the Democratic Party, rather than encouraging as many NPPs as they could to vote in the Democratic primary, kept the fact that they were allowed to do that, and how, very very quiet. And, more importantly, CA's election officials were instructed in their official training, to give NPPs ballots not containing a presidential slot, unless the NPP in question then came back and specifically asked for a special "crossover" ballot using the special magic word "crossover" that most people would be unlikely to know. Let us quote Palast:

  In some counties like Los Angeles, it's not easy for an NPP to claim their right to vote in the Democratic primary – and in other counties, nearly impossible.

Example: In Santa Rosa, Sonoma County, if you don't say the magic words, "I want a Democratic crossover ballot," you are automatically given a ballot without the presidential race. And get ready for this: if an NPP voter asks the poll worker, "How do I get to vote in the Democratic party primary?" they are instructed to reply that "NPP voters can't get Democratic ballots." They are ordered not to breathe a word that the voter can get a "crossover" ballot that includes the presidential race.

I'm not kidding. This is from the official Election Officer Training Manual page 49:

"A No Party Preference voter will need to request a crossover ballot from the Roster Index Officer. (Do not offer them a crossover ballot if they do not ask)."

They're not kidding. Poll worker Jeff Lewis filed a description of the training in an official declaration to a federal court:

Someone raised their hand and asked a follow-up question: "So, what if someone gets a nonpartisan ballot, notices it doesn't have the presidential candidates on it, and asks you where they are?
    The answer poll workers are instructed to give: "Sorry, NPP ballots don't have presidential candidates on them."
That's correct: even when people ask questions of that nature, obviously intending to vote with a party.
This affidavit, and several even more horrifying, come from Election Justice USA, a non-partisan watchdog, hoping to get injunctions to stop this nonsense. [Hear my talk with the group's spokesman, Paul Thomas, on a special edition of the The Best Democracy Money Can Buy: Elections Crime Bulletin, which I host with Dennis Bernstein on the Pacifica Radio Network.]

Let me throw in another complication. Nearly half of Californians vote by mail, ballots sent to your home automatically. Most NPP voters don't realize that, to vote in the Democratic primary, they must bring in their NPP ballot with the envelope, and say these magic words: "I want to surrender my ballot in return for a Democratic 'crossover' ballot."

Got that memorized? Because if you don't, if you say the wrong syllables, in some counties, you will be denied a Democratic presidential ballot.

...It gets far worse. There are simply not enough "crossover" ballots printed. If they run out... [according to Election Justice's filed affadavits] poll workers were told to give NPP voters "provisional" ballots even if they say the magic words "I want a crossover Democratic ballot." As I've previously reported, provisional ballots are "placebo" ballots that let you feel like you've voted, but you [often] haven't.

[Palast then noted that his KPFK co-host, Cary Harrison, an NPP, was denied voting twice in West Hollywood. He then drove to a new precinct as directed and was again denied a ballot.]

You can read Election Justice's Report about this and other techniques for biasing the primary against Sanders and for Clinton. (I am not necessarily endorsing everything this report said, but certainly it said many true and disturbing things.)

Another election-biasing ploy, unfortunately commonplace in the contemporary USA, is intentionally-biased purging. That is, in many US states, the head of election-counting (often with hire and fire power over every employee) is the "secretary of state" (SoS), often intentionally chosen to be the single most-biased person in that state. (For example, Ken Blackwell, the Bush-Cheney campaign co-chair in Ohio, was the Ohio SoS in charge of counting Ohio's 2004 votes for or against Bush.) And states must purge their voter-registration lists periodically due to registrants dying, moving away, etc. But the law often allows the SoS to select who will be purged and when, so that, e.g, Blackwell conducted pre-election purges in predominantly Democratic areas of the state.

In the case of the 2016 Democratic Primary in New York state, Clinton as former NY senator was very well connected with the entirely-Democratic-controlled NY state government, and had an excellent "machine" to deliver votes. But Sanders was actually born and raised in NY, specifically Brooklyn, so was not a complete foreigner. He lived in Vermont, a neighboring state with characteristics similar to "upstate" NY.

So what happened in NY's Dem-primary? To a good approximation, Clinton won NY City and suburbs, while Sanders won everyplace else, i.e. upstate NY, with the net result being a big win for Clinton by a 290 Kvote margin out of about 2 million cast. But, oddly enough, there were two mysterious purges of 125K voters in Brooklyn – the one borough of NY City where Sanders might a priori have been expected to have an advantage – in 2015, by the completely-Democrat-controlled NY government. NYC Mayor Bill de Blasio described "the purging of entire buildings and blocks of voters." The purge was only announced in the NY Times the day after the primary, when it became blindingly apparent that huge numbers of Brooklyn voters had been turned away. Both purges were illegal to perform at their dates under NY state law, but nevertheless occurred. This supposedly was a mere "mistake."

Also, for other mysterious reasons, polls were open for different amounts of time in different places, with the Sanders upstate areas getting fewer hours, and the Clinton NY City and nearby areas getting more hours.

To quote a pre-election announcement: "In New York city and counties of Nassau, Suffolk, Westchester, Rockland, Orange, Putnam and Erie, polls will open at 6am-9pm (EST); all other counties will have polling hours between noon and 9pm (EST)."

Well, amazingly enough, Clinton won: NY City, Nassau, Suffolk, Westchester, Rockland, Orange, and Erie, with Sanders winning only Putnam among the long-hours locations! Meanwhile among NY's 50 short-hours counties, Clinton won only 2. (And actually the situation seems even more biased than this makes it appear since, based on voter complaints, it seems the hours in reality often were more biased than the pre-planned claimed hours.)

Meanwhile, in a related (?) story, Donald Trump claimed at many times during the Republican primaries that they were "rigged." Those claims were not entirely divorced from reality. (Trump also seemed to agree the Democratic primaries were rigged, although they were not his main focus.)

Greg Palast's conspiracy theory about "CrossCheck": is discussed here. He grandiloquently claimed that was how "the election was stolen." CrossCheck was indeed a suspicious and dubious program that did look designed to bias the general election to give Trump an artificial advantage. However, Palast's numbers simply do not add up to enough of an effect to have stolen the election. At most, CrossCheck "stole" Michigan – but no other state, not even in combination with Johnson & Stein's spoiler effects.

Were Gary Johnson and/or Jill Stein "spoilers"? No.

CBS News' exit poll (apparently actually performed by Edison Research) posed the hypothetical question of who third party voters would support if the race were only Clinton and Trump. It found both Johnson and Stein supporters appeared to support Clinton over Trump by about 25 to 15 percent. But 55% of Johnson's supporters would have just sat out the election, as would 61% of Jill Stein supporters.

"The exit polling asked voters they would have cast ballots for if there were only two candidates (Clinton and Trump). A quarter of Johnson voters said Clinton, 15 percent said Trump, and 55 percent said they would not have voted. Numbers were similar for Stein voters, with about a quarter saying they would have chosen Clinton, 14 percent saying Trump, and 61 percent saying they would not have voted."
 – Stanley Feldman and Melissa Herrmann: CBS News Exit Polls: How Donald Trump won the U.S. presidency, 9 Nov. 2016.

Meanwhile according to New York Times exit polling, 63% of voters who voted for somebody other than the two major-party candidates said they would not have voted at all in a two-candidate race, with 21% going for Clinton and 16% for Trump.

Therefore, Johnson & Stein by running did hurt Clinton – and presumably did "spoil" Michigan, i.e. caused Trump to win Michigan instead of Clinton. However, their spoiler effects, even combined, were not enough to swing any other state toward Trump. And Trump still had enough electoral votes to win even if Michigan and Wisconsin both switched to Clinton.

So Johnson and Stein were not spoilers, i.e. Trump still would have defeated Clinton in a 2-way race.

Was Clinton a spoiler? Yes.

Although the press devoted a fair amount of sound and fury to the prospect that Johnson might be a spoiler, it failed to mention the far more likely possibility that Hillary Clinton was a spoiler. The decisive questions are:

  1. if Clinton (and possibly also Stein) had dropped out of the (general election) race, leaving it as a head-to-head battle between Johnson and Trump, would Johnson have won?
  2. if Clinton had dropped out shortly before the end of the primary race, causing Sanders to instead become the Democratic Party nominee, would Sanders have won?

My answers are:

  1. Yes, with probability=60%. (See below.)
  2. Yes, with probability>95%. This is based on 21 Sanders-vs-Trump pairwise polls, every one won by Sanders by margins ranging from 4% to 24%. (And note 4% far exceeds the maximum popular vote margin ever attained by an electoral vote loser.) It also is based on the known official final results, which indicated that even a slight improvement over Clinton's pairwise performance versus Trump, would have sufficed to make her win; and the fact (based on polls) that Sanders did outperform Clinton substantially (i.e. margin greater by about 8% of the number of voters) by that measure.
        The only way to dispute this is to speculate that, if Sanders had been nominated, then new attacks would have come against Sanders, which would have been much more effective in hurting him relative to Trump, than the attacks which did come against Clinton. This seems unlikely because Clinton simply had more severe-seeming attackable problems than Sanders did. There was, for example, no ongoing FBI investigation of Sanders, he never was paid over $100,000 to give a speech, and he'd had a lifelong record (unequalled by any other major rival) of refusing both PAC money and all large individual dollar contributions. As one commentator put it, Clinton was "the Queen of Wall Street, while Sanders is the cleanest major politician during the last 50 years."

In other words, not only was Clinton almost certainly a spoiler, but it indeed is more likely than not that she was a spoiler in two different ways.

Justification of answer 1. Unfortunately, "pairwise" polls about any pair besides the two major-party candidates, are rarely performed by pollsters in the contemporary USA. However, one was performed and released to the media (which of course then almost entirely ignored it) by the Johnson/Weld campaign:

Poll Question: For whom would you vote for president if the choice was between Republican Donald Trump and Libertarian Gary Johnson? (Nationwide telephone poll on 8-9 October 2016.)
   Answer             Number      Standard error (Bernoulli)
Gary Johnson:         315=41.6%   ±13.56=±1.79%
Donald Trump:         282=37.3%   ±13.30=±1.76%
Undecided/Don't know: 160=21.1%   ±11.23=±1.48%
     TOTAL            757=100%

The Johnson-Trump difference was 33 pollees out of the 757 total respondents, and this really should be regarded as 33±13.56 after putting in a ±1σ error bar, if we (rather optimistically) just use errors for one of the two candidates. And if we assume (as a worst case assumption) that the errors for both candidates are exactly anti-correlated, then 33±27.12. Consulting a table of the normal distribution, we see that this 2.43σ or 1.22σ result indicates 99.25% or 88.88% confidence that Johnson would have defeated Trump – at the time of the poll and provided the undecideds are ignored.

Also asked: "Do you have a favorable or unfavorable opinion of Gary Johnson?" Fav=15.2%, Unfav=30.0%, NoOpinion=38.1%, Never heard of GJ=16.7%. And: "For whom would you vote for president if the choice was between Democrat Hillary Clinton and Libertarian Gary Johnson?" Clinton=345, Johnson=280, Undecided=132, (Total=757).

This claim can be disputed. It is based on only one poll, and that poll was released by a biased group, the Johnson/Weld campaign. I doubt that the poll was a lie. (The reason a money-limited campaign pays for a poll is to find out the truth. And if it were a lie, then why not also try to create a more pro-Johnson impression about the "Johnson losing to Clinton" simultaneous pairwise result, and why publicize the low approval this poll found for Johnson, considerably lower than in independent public polls up til then?) However, if the Johnson/Weld campaign had conducted (say) five polls, but only released the one that made them look the best versus Trump – which seems fairly likely – then the public would get a distorted impression from that one poll, and to correct for that our "99.25% or 88.88% confidence" claims would need to be downgraded by a factor of 5 ("Bonferroni correction") to obtain (as lower bounds) 96.25% or 44.40% confidence. The former figure would mean Johnson still is a fairly good bet; the latter means Johnson-Trump was too close a pair to call.

More precisely: 100-99.25=0.75, multiply by 5 to get 3.75, and 100-3.75=96.25.

Furthermore (the disputers could continue), this poll was conducted at a comparatively unpropitious (?) time for Donald Trump: 8-9 October, shortly after the release of the "Access Hollywood" videotape in which Trump in 2005 boasted of his ability to "grab [women] by the pussy" and indeed "do anything" to them with impunity because he was a "star." This probably was not Trump's lowest point – things would seem to have gotten even worse for him somewhat later, in view of the following timeline:

Despite not being the lowest, 8-9 October likely was a comparatively low moment for Trump, plausibly lower than on election day (8 Nov). Further complicating the issue was the fact that a lot of the official voting occurred "early," namely during the period 29 Sept.-8 Nov. (The allowed early-voting periods varied depending on state but participating states typically began it about 22 Oct.)

The final grounds for disputation were that this poll pertained to the popular vote (predicting 4.4% pairwise margin for Johnson-Trump), whereas what would actually matter would be the electoral vote. Historically, the popular and electoral winners have coincided over 90% of the time. And the whole judgment ignores the "undecideds," whose number was substantial.

My bottom line: pending any further pairwise poll data (since at present I know of no other such poll), my personal estimate is that there was about a 60% chance that Clinton was a spoiler since Johnson would have defeated Trump head to head.

If so, this poll plus the official election results also would imply the existence of a Condorcet cycle

Johnson > Trump > Clinton > Johnson

where "A>B" means "A would defeat B in a head-to-head race using official system."

Not only that, it would show that Clinton not only was a spoiler with the official election method, but also was an IRV spoiler, i.e. by running, Clinton prevented Johnson's victory over Trump using instant runoff voting.

In other words, anti-Trump pro-Clinton voters, by honestly voting Clinton-top, would with instant runoff voting (as well as with the official actually-used voting system) have made a strategic mistake. They would have been better off dishonestly voting Johnson top, because that would have caused a better election result in their view.
    It is often contended – of course utterly falsely – by instant runoff proponents that IRV "eliminates" the spoiler effect and the "wasted vote" problem, or somehow weakens 2-party domination permitting "third parties" to win. This 2016 election provides yet another refutation, perhaps the most prominent, of those false claims.

Instant Runoff Voting (IRV) would not have altered the winner

And of course Trump still would have won the general election with instant runoff voting conducted either within states, or within the electoral college. However with a straight nationwide vote (no electoral college) Clinton would have won with IRV.

These facts are proven by the official election results combined with the CBS News exit poll's question about "second choices" of third-party voters.

The only possible way to dispute this would be to hope that strategic lying on the official (plurality style) ballots was so vast, that actually with instant runoff voting, the voters would have been way more honest and less naively-strategic, causing Clinton and Trump not anymore to reach the "final two." However, that hope is quashed by the Civis online poll, which actually did ask their respondents to provide rank-order ballots, and indeed did find that Trump and Clinton were the final two, with 43.8% ranking Clinton top and 43.8% ranking Trump top. (Civis also approximately confirmed the "25 to 15" finding about Trump/Clinton preferences of Johnson and Stein voters from the CBS exit poll.) The difference between 43.8% and the 10.4% who ranked anybody besides Clinton and Trump top is so great that, despite issues I have with the quality of the Civis poll, this particular claim seems indisputable.

This election was a huge distortion of democracy

Obviously, those pairwise results for Johnson-vs-Clinton and Johnson-vs-Trump demonstrate that Johnson, in truth, was vastly more supported by the American public than one would suspect from his 3.3% sliver of the official plurality vote. (Which also is independently demonstrated by the approval- and score-voting polls that we shall discuss soon.) Indeed, it even is arguable that Johnson actually should have won the presidency.

This enormous distortion is caused by strategic voting imperatives ("must not 'waste my vote' by voting for anybody besides the leading two") which in the earlier USA 2000 election were known from NES data to have caused 90% of honestly Nader- or Buchanan-favoring voters to vote for somebody else. This is a well known defect of the USA's abysmal plurality voting system.

A different way to see that strategic distortion at work: The top 7 states for Gary Johnson (highest 7 percentages of popular vote) were, in descending order,

NM (9.3%), ND (6.2%), AK (5.9%), OK (5.8%), MN (5.7%), SD (5.6%), WY (5.2%).

All those states were "easy call" states in which the |Trump-Clinton| margin ranged from 8.2% (NM) up to over 45% of the vote (WY). Meanwhile, in the 8 closest states – those with |Trump-Clinton| margins below 3% – Johnson always got ≤5.1%.

That is presumably because those states' voters knew that voting for Johnson incurred extremely tiny risk of causing a "spoiler" scenario. There is one exception: The 8th-most pro-Johnson state, Maine (ME), which voted 5.1% for Johnson, actually did have a small |Trump-Clinton| margin (2.7%), but Maine is one of the two states that awards its electoral votes by congressional district rather than statewide, and those districts are "gerrymandered," so again its voters knew they had low spoiler risk.

Similarly, but more clearly: the top 17 states (regarding DC as a "state" for this purpose) most-supporting Jill Stein, all were extremely "easy call" states with |Trump-Clinton| margins ranging between 12% and 85% – plus Maine:


This extra clarity was perhaps because Green voters were more cautious than Libertarian voters due to their previous bad experience with Ralph Nader as a Green Party spoiler in 2000.

Yet another signature of the huge distortion: An online post-election poll by Civis Analytics (1084 registered voters who said they'd voted, Nov. 15-16, poll commissioned by found that the percentages of voters who "found [candidate] acceptable or somewhat acceptable," and who would have voted for that candidate in a head-to-head race versus Clinton, were respectively

Candidate Acceptable or Somewhat Preferred over Clinton
Hillary Clinton48.4
Donald Trump46.749.3
Gary Johnson32.445.0
Jill Stein29.042.1
Evan McMullin25.939.1
Darrell Castle17.637.5

These percentages are vastly greater than anybody would have naively guessed from these candidates' official vote shares.

[But I warn the reader that the Civis poll because based on a sample of online voters with unstated "demographic correction factors," is of less reliability than genuinely random telephone polls conducted by established pollsters who publish full polling reports. My requests to Civis for their raw data, fuller data, or a report, all were ignored. Despite those problems, I think the effects we just reported are so enormous that they cannot be disputed.]

Enormous costs incurred by USA's foolish use of plurality voting system

The enormous costs the USA (and hence world) are paying for foolishly using that system were quite clearly demonstrated yet again by this 2016 election. Here is a list of features of Trump/2016 – most or all of which were bad – which set new records:

  1. On election, Trump with 37 or 38% approval was the least-approved president (during the era of approval-style polling, i.e. starting in the late 1940s, at the date of first assuming office, except in Trump's case we use election day). Furthermore, Clinton, if she had won, also would have been the least-approved ever, with 41 or 42% approval. The previous low was Gerald Ford with 47% initial approval, as we see in this graphic from Gallup. (Also Jimmy Carter and Harry Truman jointly held the record for lowest approval time-averaged over their entire terms, at about 45.5% each. Since evidently approval usually decreases during presidential terms we have good a priori reason to suspect Trump will be one of the worst presidents in the last 100 years.) Gallup presidential approval versus time plot
    Furthermore, according to several pollsters using various approval- and score-based measures, Trump was the least-wanted among all his historical rivals (i.e. major-party presidential nominees of the past) at the same month in their campaigns, during April, May, June and July – and Clinton also would have held that record all those months too if it were not for Trump. (Example, another, another.) The previous worst had been Barry Goldwater in 1964.
        Update: Additional post-election polls found that, e.g, Trump approval was 37% favorable and 51% unfavorable in a Quinnipiac nationwide telephone poll of 899 voters nationwide on 5-9 January 2017. So there is no question Trump was the least-approved president (on day 1) ever recorded.
        Meanwhile, the public's job-approval rating for congress was also abysmal. The all-time low according to Gallup during 1974-2016, was 9% approval, achieved in Nov. 2013. By early Nov. 2015, congress had improved to 11% approval and 86% disapproval. (The highest ever had been 84% approval in October 2001.) This was probably due to a Republican policy of total obstructionism in both branches of congress throughout the Obama administration, combined with several times threatening to shut down the entire government as a blackmail tactic.
  2. The 2016 election had the highest-ever-recorded "gender gap": 13 points:
    YearWomenMenDifference Exit Poll Source
    197238 36 2CBS
    197650 50 0CBS
    198045 36 9CBS/NY Times
    198444 37 7CBS/NY Times
    198849 41 8CBS/NY Times
    199245 41 4Voter News Service
    199654 4311Voter News Service
    200054 4212Voter News Service
    200451 44 7National Election Pool
    200856 49 7National Election Pool
    201255 4510National Election Pool
    201654 4113 CNN exit polls
    Percentages of women and men voting for Democratic US presidential candidate.
    However, the sexism combined with racism (?) in the sense that 52% of white women voted for Trump.
  3. Trump was, by a goodly lead, the lyingest major presidential candidate the USA ever had (during the era of fact-checking organizations, i.e. starting 2003; "major" here meaning "anybody looking ≥15% likely to win one of the top-two-parties' nominations"), as reckoned unanimously by all of the USA's top three independent fact-checking entities,, and Washington Post fact checker. Here is Trump's (astounding) politifact report card as of election day 2016:
    TRUE           14 (4%)
    MOSTLY TRUE    37 (11%)
    HALF TRUE      49 (15%)
    MOSTLY FALSE   63 (19%)
    FALSE         111 (34%)
    PANTS ON FIRE  57 (17%)
    This was based on Politifact's checks of 331 statements by Trump; 70% of his statements fell in their mostly-false, false, or pants-on-fire categories! And here is the Washington Post's independent fact-checking summary of H.Clinton and Trump as of 3 Nov. 2016:
  4. With 24 women publicly accusing Trump of sex-assaults or other sexual misconduct, that surely also was a record.
  5. Trump, at 70, was the oldest ever to ascend to the US presidency. But fear not – despite his age, Trump's physician provided a letter to the press "stating unequivocally" that Trump would be "the healthiest individual ever elected to the presidency."
  6. Trump was the least qualified US president ever, in the sense that he was the only one with no previous government (civilian or military) job experience. (In contrast, Hillary Clinton's extensive experiences would have made her one of the most qualified presidents ever, while Gary Johnson was former governor of New Mexico as well as a wealthy self-made businessman.)
        Trump supporters could console themselves with the fact that his VP Mike Pence (governor of Indiana 2013-2016, and US congressman 2001-2013) had 15 years of experience. Unfortunately during his 12 years as congressman, while Pence wrote 90 bills and resolutions, he failed to get even a single one to pass into law.
  7. Trump was the most litigious US president ever, as is detailed in USA Today's interactive database Three decades, 4095 lawsuits. (And probably there are more than 4095 based on the fact that USA Today's tally kept growing as they kept finding more, for many months. Their initial tally in June 2016 was only about 3500. Also, Trump threatened to sue people far more often than he actually did – probably tens of thousands of times.) Hundreds of the lawsuits were the result of Trump refusing to pay people he'd done business with. Even if Trump spent only 1 day per lawsuit, excluding weekends and holidays, that would mean at least 16 out of those 30 years of his life consisted of him either suing or being sued! Probably that would be physically impossible for almost anybody else to accomplish, even if they tried. At least 75 of the lawsuits still are in progress as Trump enters the Oval Office, most notably cases alleging his "Trump University" was a fraudulent "get rich quick" scam...
        ...and one of the three rape cases. (The other two rapes were settled. The third, still ongoing, case alleges both Trump and billionaire and convicted/jailed sex offender Jeffrey Epstein in 1994 both repeatedly and collaboratively raped a 13-year-old girl – the plaintiff, called "Jane Doe" for mysterious legal purposes in the case. The lawsuit includes a witness of 4 of those rapes by Trump, which means it should be taken more seriously than most rape cases, because witnesses of rapes are rare.) Update: The Trump university case was settled after Trump agreed to pay $25 million in damages, thus refunding about 50% of the tuition paid by the 6000 plaintiffs.
  8. Trump was the most married & divorced president ever, with 3 marriages and 2 divorces (so far). They always are younger than him and keep getting younger:
           Wife           Her birthyear    Country    Married years   
    Ivana Marie Zelnickova       1949   Czechoslovakia   1977-1992
    Marla Maples                 1963   USA (Georgia)    1993-1999
    Melania Knavs                1970   Slovenia         2005-
    Only one preceding president divorced even once: Ronald Reagan. That presumably is because presidents tend to have excellent "people/relationship skills," good career success, and high attractiveness, and as a side effect their marriages succeed more often than most people's.
  9. Trump as far as I can tell was the only major presidential candidate ever to publicly call for a foreign power to spy on the USA, and in particular on his political opponent:
    Russia, if you're listening, I hope you're able to find the 30,000 emails that are missing [from emails that Hillary Clinton turned over to the State Department]. I think you will probably be rewarded mightily by our press. Let's see if that happens. That will be next. Yes, sir.
    – Trump in a lengthy news conference in Doral, Florida [Ashley Parker & David E. Sanger: Donald Trump Calls on Russia to Find Hillary Clinton's Missing Emails New York Times 27 July 2016; electronic version includes video of Trump saying it.]
    Also, if it is correct (as many US intelligence agencies claimed) that Russian intelligence made a concerted effort to influence this election in Trump's favor, then (a) in view of how close the election was, that effort likely was successful, and (b) this marks the first time in US history that any such attempt was successful.
  10. Some have argued that a better system than "democracy" would be "epistocracy" (a coined word meaning "rule by the knowledgeable") in which, only "knowledgeable" people would be allowed to vote, or more knowledge would give your vote greater weight, or some such. I am dubious epistocracy could be safely accomplished. But regardless of that debate, this election did provide some experiments about precisely that question. Namely, a set of people unquestionably more-knowledgeable than the average voter, were the editorial boards of major US newspapers. Here is a tally by wikipedia of the endorsements made by the USA's 100 top (reckoned by paid circulations) newspapers:
    #newspapers2016 presidential endorsement
    57Hillary Clinton (Democ)
    4Gary Johnson (Liber)
    3Not Trump (anti-endorsement!)
    2Donald Trump (Repub)
    34(No endorsement made)
    (This compares with the preceding 2012 election, in which the major newspaper endorsements split approximately 51/48/1 percent for Obama/Romney/Johnson.) So it seems likely that 2016 was the single most-lopsidedly-bad US election ever, reckoned by major newspaper endorsements; and this "epistocracy" differed quite drastically from the actual democracy.
        A different set of people more knowledgeable than average about Trump were the residents of Manhattan where Trump (who lived there, in the penthouse of "Trump Tower," since its construction ended in 1983) had been a major figure for 35 years. Manhattan's vote was: Clinton 515481, Trump 58935, approximately a 9:1 ratio. (Plus Stein and Johnson each got about 1%.)
        Finally, another set of people more knowledgeable than average were former presidents. All former presidents still alive at the time of this election (Obama, Bush II, W.Clinton, Bush I, Carter) unanimously said – either to the press or to multiple sources who told the press – they were voting for Clinton, with the exception of Bush II, who told USA Today he'd intentionally not voted for President.
  11. Trump's bigtime public political career began in March 2011 with his trumpeting (e.g. on television) of the "birther conspiracy." This was the theory that then-sitting President Obama was not really born in Hawaii USA, because his birth certificates were fraudulent – he'd somehow been brought as a young child to the USA by evil forces who'd pre-programmed him to become an evil president? – and therefore Obama actually was ineligible to be president. Trump was by far the loudest voice for the "Birthers." He claimed he'd sent investigators to Hawaii to research the matter ("and they can't believe what they're finding!"), but never said who they were or what they found. Finally in September 2016, Trump for the first time publicly acknowledged that Obama was born in the USA – then falsely claimed the rumors to the contrary had been started by Hillary Clinton in 2008! [Maggie Haberman & Alan Rappeport: Trump Drops False 'Birther' Theory, but Floats a New One: Clinton Started It, New York Times 16 Sept. 2016.] This is probably the only time in world history that a prominent national leader publicly disavowed his own signature career-commencing act, and then attributed it to his rival (who of course had been attacking and ridiculing it all along).
  12. Trump was the only major presidential candidate since the 1960s to refuse to release his taxes.
  13. Trump may or may not have been the wealthiest US President. The three richest were D.J.Trump, J.F.Kennedy, and George Washington; but I am unable to decide the correct ordering among them. Trump thoughout his life greatly exaggerated his wealth.
        With economic inequality at or nearing record-high levels in the 2016 USA (mainly due to the rich effectively compounding their money faster than the rest of us, a trend which started with Reagan's great lowering of high-end tax rates in the early 1980s, and probably helped by the onset of "globalization") it seems unlikely that Trump's proposals for further lowering of high end and corporate tax rates – apparently designed to enhance his personal financial situation? – are going to improve that.
        Trump's actual wealth is extremely unclear because:
    1. He refused to release his taxes.
    2. The contracts and loans he signs often were kept secret.
    3. Many of "his" holdings actually are partly or entirely owned by others. For example near the end of his Presidential campaign (4 Nov.) the "Trump International Hotel & Tower" in Toronto – the tallest residential building in Canada – was placed into receivership just 4 years after Trump and his children conducted its opening ceremony. Yet another bankruptcy for The Donald? Actually not, since according to press reports his company had not built it and his ownership share was zero! The Trump Organization merely managed its operations, plus ran various misleading promotions which made it seem like the developer/owner. That all turned out to be bad news for the actual owners.
    4. Trump's business is a privately held corporation – in contrast to publicly held, i.e. stock-traded, corporations, which have legal disclosure requirements.
    Actually Trump, in 1995 when presumably in severe financial trouble, did create a stock-traded corporation "DJT," raising $140 million in its initial public offering. He then talked it up and made things look promising, getting its stock price to multiply 2.5× in 1 year. At that point, according to Fortune magazine, "Trump's 41% stake [was] worth around $400 million, [which] was perhaps the only time a major pillar of [Trump's] net worth was verifiable and transparent." Then Trump switched to basically acting as a "corporate raider" on his own corporation (he had complete control as chairman of its board for its entire existence, CEO for about half, and the stock he owned was of a special class unavailable to others that carried many more votes), sucking its assets and market value out into his own pocket via a variety of absurd transactions with other Trump-owned companies (while leaving its debt in place), plus paying himself enormous bonuses for his "excellent performance" – albeit causing its investors to lose 90 cents out of each invested dollar by early 2005 after it declared bankruptcy with a net declared loss of $647 million (it had lost money every year). The investors were virtually wiped out, but it was an excellent killing for The Donald. Two plots of DJT stock performance As you can see from the graphics, DJT vastly underperformed both the stock market as a whole, and its casino-stock segment. Indeed, it was by far the single worst performing among all casino stocks. Aggregated over the 3 years 2000, 2001, and 2002 (observe DJT's "excellent" performance during that period on the lefthand graphic), DJT paid Trump $4.25 million in salaries, $8.7 million in bonuses and/or "other compensation," and $1.5 million in stock options.
        After DJT's demise, Trump immediately did it again, in December 2004 creating TRMP, his second publicly traded corporation (consisting largely of remnants of DJT), whose fate was similar. At one campaign speech I witnessed, Trump waved an official-looking piece of paper, and said it was his latest "net worth" which "just came in" and read $12 billion. Meanwhile Forbes Magazine in 2016 estimated it at $3.7 billion, Fortune at $3.9 billion, and Bloomberg News at $3.0 billion. At the other end of the scale, Trump's biographer, the financial journalist David Cay Johnston, estimated Trump actually was worth below $1 billion.
    Johnston wrote the 2016 book The making of Donald Trump which recounts many stories of how Trump exaggerated his wealth and about his illegal, unethical, and/or unsavory business practices. Those caused Trump's fellow New York zillionaire and former Mayor Michael Bloomberg, estimated to be worth $42.7 billion, to comment "if Trump wants to run the nation like he's run his business, God help us!" A larger biography by two Washington Post journalists is Michael Kranish & Marc Fisher: Trump Revealed: An American Journey of Ambition, Ego, Money, and Power (2016); and a personal recounting of Jack O'Donnell's experiences working for Trump as his casino manager is Trumped! (1991). O'Donnell paints Trump as an irresponsible mismanager, a huge narcissist and braggart, the biggest gambler he ever encountered, a philanderer, and a racist.
  14. But Trump definitely does hold this record: Trump is the president who had, at one point, the largest negative wealth, after enormous bankruptcies. Indeed Trump may not only hold this record among US presidents, but is in a small group of contenders for the world record as the person, in all world history, with the greatest personal debt in bankruptcies! In a nutshell the situation was this: Trump borrowed heavily, especially to build/buy huge casinos in Atlantic City, New Jersey. For example he financed his "Trump Taj Mahal" casino in 1990 with "junk bonds" at an extraordinarily high 14% interest rate. After it went under in 1991, the enterprise was $3 billion in debt. Trump also personally guaranteed a lot of his debts, i.e using his own personal wealth as collateral, and reusing that same wealth as more collateral to get different loans from different lenders (who did not know about the previous loans, since Trump kept their terms private). Then in 1992 his "Trump Castle" (opened in 1985) filed for bankruptcy. And then again in 1992 his also-huge Atlantic City property, the "Trump Plaza Hotel and Casino," was crushed beneath $550 million in debt landing Trump in bankruptcy court a third time. Then again in 1992, he filed for bankruptcy on the "Trump Plaza" hotel in New York City. About that time, Trump later told The Washington Post, he passed a beggar in New York and told his then-wife, model Marla Maples, "You see that man? Right now he's worth $900 million more than me." [David S. Hilzenrath and Michelle Singletary: Trump went broke, but stayed on top, Washington Post 29 Nov. 1992.]
        As a result, Trump should have been personally bankrupt on an enormous scale. The only reason that did not happen is the old story that "if you default on a $20000 loan, that's your problem. If it's $500 million, that's the bank's problem." Trump's creditors held a conference and agreed (after rancorous internal disputes) to accept pennies on the dollar, take over ownership and management control of those casinos, and allow Trump to financially survive, as opposed to completely wiping him out, in the hope that he might be able to recover some of their money later. But the damage continued. In 2004, Trump faced his fifth corporate bankruptcy, when "Trump Hotels and Casinos Resorts" became buried under $1.8 billion in debt. Then in 2009, "Trump Entertainment Resorts Inc." (TRMP) became his sixth corporate bankruptcy after missing a $53 million bond interest payment. After its $4 share price plunged to about 25 cents it was delisted from the NASDAQ. (As of 2016 some online financial information sites still list it with share price apparently frozen at about 6 cents, with trading volume zero.) After some recovery it declared bankruptcy yet again in 2014 (which perhaps should count as Trump's seventh bankruptcy?) and continues on, but now as a subsidiary of Icahn Enterprises and with ticker symbol now "TER."
        So evidently, billions of dollars in loans to Trump and his Organization were never paid back. That amounts to a major component of his estimated total wealth. Indeed if Johnston's estimate that Trump's wealth is below $1 billion is correct, then Trump actually got substantially more money in never-repaid loans, than his total fortune.
  15. According to the Washington Post, Trump holds the record, by far, as the US president with the largest and most conflicts of interest. (Update: in a news conference 11 January 2017, Trump refused either to sell his businesses or to place his assets in a "blind trust" while serving as president, which would seem to bring him into immediate enormous conflict with the "emoluments clause" of the US constitution and also contradicting numerous vague claims he'd made during his campaign. Trump also appointed his son-in-law Jared Kushner as "senior advisor," seeming to bring him into immediate conflict with anti-nepotism act of 1967.) One problem was that, after his huge casino-bankruptcies, essentially all US banks refused ever to loan Trump any money again. He however was able to obtain financing from foreign banks – which are known to include (the highly corrupt) Deutsche Bank, several Japanese banks, and the National Bank of China, albeit with Trump continuing to keep his finances private it is difficult to see the whole story – although supposedly only under very restrictive terms which give those banks unusually large amounts of power over him. Also, Trump's son Eric claimed during the 2016 campaign that the Trumps' major construction projects were now mainly in Russia; he is also known to have projects in India, Azerbaijan, Saudi Arabia, United Arab Emirates, Turkey, Brazil, Argentina, Malaysia, and China. Trump also, as election day approached, had an upcoming labor dispute in Las Vegas, to be mediated by the NLRB that he (as president) would appoint; the anti-Trump side could be argued by lawyers from the Justice Department, whose head also would be appointed by Trump. Trump's taxes are being audited by the IRS, whose head Trump will appoint. Trump also had just finished building a large hotel on land he rents from the US government (GSA) in Washington DC. Trump will appoint the head of the GSA. Diplomats from a large number of (essentially all?) foreign countries are now supposedly making plans to lodge at that Trump hotel in an effort to bribe him. [Drew Harwell: Trump's conflicts of interest are without precedent in American presidential history, 9 Nov. 2016; Sam Thielman: Trump's conflicts of interest take White House into uncharted territory, The Guardian, 12 Nov. 2016.] Trump also has simply directly held many presidency-related events at Trump hotels and resorts, billing all attendees, the Secret Service protecting him, etc, for their attendance, which typically earns him around $500K per event. President-elect Trump according to press reports (and a later ambiguous response by himself) spoke with British UKIP-leader Nigel Farage about halting plans for a wind-energy farm near his Turnberry golf resort, because it would mar the view. According to a report by prominent Argentine journalist Jorge Lanata, Trump's first post-election phone call with his Argentine counterpart Mauricio Macri included urging the latter to steamroll permit issues holding up construction of a new Trump-brand skyscraper in Buenos Aires. Both Macri and Trump quickly denied the report. [Revelan que Donald Trump le pidio permiso a Mauricio Macri para construir su torre en la Ciudad, La Nacion, 20 Nov 2016.] Late in October, Phillipine president Rodrigo Duterte appointed Jose Antonio, a longtime business associate of Trump's, as a "special envoy to the United States." [Richard C. Paddock: Trump's business partner is new Trade Envoy to U.S, NY Times 9 Nov 2016.] In December, Trump's main business partner Dogan on the "Trump Towers" project in Istanbul was arrested in an apparent open-ended effort by Turkish President Erdogan to coerce Trump. (Erdogan had earlier, in June, called for Trump's name to be removed from those towers.) Trump appoints the heads of the agencies that investigate/regulate the banks he has huge loans from. Setya Novanto happened to be standing next to Trump when Trump announced on US television (1 Sept. 2016) from Trump Tower, his signing of a "loyalty pledge" to the Republican Party. At that announcement Trump described him to the cameras as "the Speaker of the House of Malaysia" and a "great man" who was there with "his whole group" to meet Trump to "do great things for the United States." However, what Trump neglected to mention was that Novanto had been forced to resign as Speaker in December 2015 due to his part in an enormous corruption scandal attempting to extort Indonesia's largest taxpayer, U.S. mining giant Freeport, to pay him $4 billion in bribes. (He was secretly recorded doing so.) Also as part of the same or related scandals, global investigators believe more than $1 billion entered Malaysian Prime Minister Najib Razak's personal bank accounts. And, amazingly enough, the man Trump had touted throughout his early campaign as his greatest friend and future US Treasury Secretary – Wall Street barracuda Carl Icahn – was the largest owner of Freeport! Trump's son-in-law Jared Kushner – also the scion of an ultra-wealthy New York City real estate family (his father and business partner Charles in 2005 was sentenced to two years prison for tax evasion, witness tampering, and illegal campaign contributions) – also has enormous but largely secret financial entanglements with Russians and Chinese.
  16. Trump, as one of the central facets of his campaign, pledged to build one of the largest walls ever constructed along the USA-Mexico border, plus forcibly deport (with the aid of a new military "deportation force") illegal immigrants estimated to number about 11 million, which would be one of the largest forced migrations in human history:
    Large forced migrations. All numbers approximate.
    Who forced it     When      Who moved                How many    Deaths
    Slavers        1500s-1800s   Africans                  20M         3M
    USA               1800-1900  Native American Tribes    200K?       10K?
    A.Hitler+Nazis    1940s      Jews ("holocaust")         6M         most
    A.Hitler+Nazis    1940s      Non-Jews ("holocaust")     5M         most
    A.Hitler+Nazis    1939-1943  Poles forced to resettle   1M         ?
    J.Stalin          1930-1     Kulaks                    1.8M        500K
    J.Stalin          1940s   many selected ethnic groups  3.3M        1.5M
    W.Churchill       1940s  Ethnic Germans in Poland, 
                                  Hungary, Czechoslovakia  10M         3K
    Yugoslav & Romanian govts  More ethnic Germans (expelled) 12M      ?
    Rwandan war       1994       Rwandans                  2.4M        650K
    Khmer Rouge       1970s      Cambodians                1M?         2M
  17. Trump never held a press conference in 2016 after the one on 27 July where he'd urged Russia to hack Hillary's emails. (Indeed his next one was 11 January 2017, about half a year later.) Meanwhile, Hillary Clinton failed to hold a press conference throughout a 277-day span between 5 Dec. 2015 and 8 Sept. 2016, with the arguable possible exception of taking some questions during a minorities conference on 5 August.
  18. This may have been the first USA presidential election swung by "fake news." Well... certainly, lies have, historically, often been circulated during campaigns. But in the present case with the rise of the internet and "social media" such as "FaceBook," the volume of "fake news" stories circulated became massive to the point where they actually could overwhelm genuine news. Many appeared to the eye to have been written by genuine reputable news agencies such as ABC News (complete with official ABC News logo), etc. But actually, ABC had nothing to do with it and the story was a complete fabrication. An analysis by BuzzFeed [Craig Silverman: This Analysis Shows How Fake Election News Stories Outperformed Real News On Facebook, 16 Nov. 2016] showed that – stunningly – FaceBook news feed's top 20 genuine news stories related to the election were outperformed by its top 20 election-related fake news stories, during the period from 1 August to Election day!
        Was this enough to swing the election? Certainly. First, Facebook claimed more than 1.65 billion monthly active users as of 31 March 2016, far exceeding the USA's electorate, and claimed those 20 fake news stories got 8.7 million "shares, reactions, and comments" as measured by FaceBook – the plain readership (i.e. readers who just read it without "sharing" or "commenting") presumably was larger. Second, BuzzFeed found that of these top 20 fake news stories, 17 were overtly pro-Donald Trump or anti-Hillary Clinton! That means, unquestionably, that they were the result of a conspiracy. Third, consider how close the election was. If less than 0.1% of FaceBook users, or just 6% of those 8.7 million sharers, reactors, or commentors, switched their vote to Trump due to this, that would have been enough. And keep in mind that FaceBook was only one fake-news venue.
  19. This one seems a bit ominous. According to a compilation by wikipedia 2013-2014 data; "Measure of America" calculations using mortality data from the Centers for Disease Control and Prevention, National Center for Health Statistics 2009, and population data from the CDC WONDER database, here are the best 9 US states for life expectancy at birth, and (on the right) the worst 9 states.
    StateYears StateYears
    Hawaii 81.3S.Carolina 77.0
    Minnesota 81.1Tennessee 76.3
    Connecticut80.8Kentucky 76.0
    California 80.8Arkansas 76.0
    Massachusetts80.5Oklahoma 75.9
    New York 80.5Louisiana 75.7
    Vermont 80.5Alabama75.4
    New Hampshire80.3W.Virginia 75.4
    New Jersey 80.3Mississippi 75.0
    All 9 states on the left voted for Hillary Clinton. All 9 on the right voted for Donald Trump.
  20. And incredibly, a US state statistic perhaps even more correlated with Clinton/Trump election result was each state's percentage of college graduates.
    Massachusetts (40.5%), Colorado, Maryland, Connecticut, New Jersey, Virginia, Vermont, New Hampshire, New York, Minnesota, Washington, Illinois, Rhode Island, California (31.4%) West Virginia (19.2%), Mississippi, Arkansas, Kentucky, Louisiana, Nevada, Alabama, Indiana, Oklahoma, Tennessee, Wyoming, South Carolina, Idaho, Ohio (26.1%)
    On the left are the 12 most-educated US states as reckoned by the percentages (ranging from 31-41%) of their population with a bachelor degree or above during 2011-2015. Also, Washington DC has 54.6% bachelaureates, surpassing any state. All 14, plus DC, voted for Clinton. On the right are the 14 least-educated US states by the same metric (19-26% bachelaureates). All of them except Nevada voted for Trump. [Source: US census.] Why did Nevada break the pattern? Perhaps it has something to do with the fact that both Nevada and Trump had great experience with gambling casinos.

Will Trump be a great president? I don't know. But all those simultaneous worst-ever records are not a good start... for most of them just that one would seem under normal circumstances to make him unacceptable as President.

Incidentally, one could construct a (smaller) list of often-bad news about Hillary Clinton too. The most obvious superlative about Clinton is that she if elected would have been the first female president (no woman had ever been vice president either, for that matter).

After her husband Bill's presidential terms ended, the Clintons claimed they were virtually broke, but then Hillary made over $150 million from giving speeches to Wall Street firms for $200K to $700K per speech (text of speeches kept secret and recording forbidden by special contract), while Bill got at least $17.6 million for being "honorary chancellor" 2010-2015 of a chain of for-profit "Laureate universities" (when their founder Doug Becker was asked what Clinton did in that "job," he responded that he'd served as "an inspiration"). Bill Clinton also got huge payments for making speeches, e.g. in May 2012 earned $1.4 million for a 7-day-long tour giving speeches to rich people's groups in Europe, and in all earned over $100 million for giving speeches during 2001-2013, most of it from foreign countries (many while his wife served as US Secretary of State 2009-2013). In contrast, the annual salary of the US president during 2001-2016 was $400K, while the USA per capita income was about $40K, so Hillary actually often earned more for a single speech to a Wall Street Firm, than she would have gotten for an entire year's service as US president, or 10 years earnings for an average citizen.

I don't know about you, but I simply cannot regard those as "perfectly normal legal economic transactions." They are vastly outside the norm. The only way I can interpret them is this: the vast majority of the Clintons' total fortune was acquired 100% corruptly. Considering a major problem faced by the 2016 USA was its record or near-record high levels of economic inequality, it did not seem wise to elect somebody that severely bought and paid for as the "cure." Also, as Bernie Sanders' campaign (following, e.g, an exposé by the Washington Post) pointed out, Clinton cheated about campaign finance laws, using shenanigans resembling the game of "3-card monte" to effectively obtain unlimited individual campaign donations despite legal limits on donation amounts.

Was Trump's victory a "huge surprise"? Not to me and some others.

Many in the media proclaimed that it was a "huge surprise" which "nobody predicted" when Trump won the Republican nomination, and then when he won the presidency, they said it was the "biggest upset in living memory."


Well, first of all, I myself publicly predicted, back in July 2015, that Trump should easily win the GOP nomination as a consequence of "fame based failure" historically-confirmed plurality voting dynamics. (Which, by the way, Trump himself may well have been fully aware of. At least, he certainly acted like he was.) So while it may be true that major media-approved pundits using their usual no-evidence gut-based "methodologies," did not predict this, it is false to claim "nobody" did. It is merely that the media was not interested in what I had to say.

I further went on (back in July 2015) to predict that

  1. Sanders would be an underdog versus Hillary Clinton to win the Democratic nomination, but would provide a tough battle to a fairly close finish.
  2. Trump would then be an underdog versus Hillary Clinton in the general election, but still would have a "puncher's chance" to win – likely depending on scandals, i.e. which one got more damaged by various possible scandals breaking.
All of those predictions by me turned out, 1.5 years later, to be correct. What I did not predict – but could have if I'd been a bit cleverer – was that Trump would, toward the end of the GOP primary battle, encourage anti-Trump voters to stop voting for all kinds of possible Trump rivals, and instead focus on Cruz alone – because in his view, Trump and Cruz were the only two left with serious chances, and any other vote would therefore be "wasted." (This went against Trump's successful strategy all along of splitting the vote between all his rivals.)

The reason that (I believe) Trump did that was: At that point, there was no longer any serious danger that Trump would not win the plurality of GOP delegates. The main danger was that he would fail to win a majority, and then all the anti-Trump delegates might unite to nominate somebody else to save us from Trump. (There were, in fact, attempts to organize exactly that, one of the ringleaders being the previous GOP nominee Mitt Romney.) If Cruz came in a strong second behind Trump, then it would be difficult for that conspiracy to justify choosing anybody besides Cruz; and as Trump well knew, the Republican Establishment detested Cruz.

The second thing I did not see coming was the Trump sex-assault/videotape scandal. Fortunately for Trump, Hillary Clinton and the DNC had countervailing email scandals (which everybody did know was looming – Sanders had declined to exploit it during the primary, but obviously the Republicans and FBI were not likely to be that chivalrous) which happened to get re-emphasized by the media during the 2 weeks immediately preceding the election.

Numerous polls in the timespan leading up to the election usually had Clinton ahead of Trump, and consequently poll-aggregators such as HuffPost Pollster and predicted (correctly) that Clinton would win the national popular vote, by (respectively) 5.3% and 3.3% margins. Actually, she won by a 2% margin. Given that these two aggregators differed by 2, one might expect each of their predictions to exhibit standard error of about ±2, in which case the actual margin was not outrageously off target.

But the electoral vote was what officially mattered, and that is less trivial to predict – statewide, not nationwide, polls are needed, and the individual state-predictions need to be combined, e.g. using "Monte Carlo" computerized methodology. The well known prediction agency "five-thirty-eight" run by statistician Nate Silver thus predicted (the day before the election) a 71.4% chance that Clinton would win, versus 28.6% chance for Trump. Meanwhile the PredictWise "prediction market" forecast 89% chance for Clinton. In view of that, it perhaps was a "surprise" that Trump won, but hardly a huge one. The amount of surprise was comparable to: if you roll a 6-sided die, and the result is "1" as opposed to a member of {2,3,4,5,6}. I've performed this experiment, often with result=1, many times in my life and its "surprise" value wore off long ago.

So I am not at all impressed by the media's expression of huge surprise.

And now for something completely different: Lichtman's "13 keys" to the white house

History professor Allan J. Lichtman (American University) in the Washington Post on 23 September 2016, publicly predicted Trump would win the presidency. This was based on 13 particular yes/no questions, which he called for short, the "keys." The 13 keys were devised by Lichtman in collaboration with Russian scientist Volodia Keilis-Borok in 1981. To use them, you simply answer the 13 questions, and if more than half the keys are true, then the incumbent party is predicted to retain power, while if more than half are false, the challenging party will win the White House. Lichtman told the Washington Post that this method had successfully predicted all 8 presidential elections between 1983 and 2013 – and now with 2016 added, that makes 9.

The interesting thing about Lichtman's "13 keys" method is that his 13 questions only concern the historical circumstances surrounding the election, not the election itself. It, roughly speaking, does not care who the candidates are, what they say, and what their poll numbers are. For that reason it often is possible to produce their prediction years ahead of time. In particular: if the Republican nominee had not been "Donald Trump," but rather "Porky the Pig" – and Porky was polling only 1% – Lichtman's prediction would remain unaffected!

As an perhaps-unspoken corollary, Lichtman is telling us that voters are complete morons! They are wholy predictable robots. They do not even look at what the candidates are saying, what they will do once elected, what their records are, etc. They simply mindlessly choose the party appropriate for that historical moment, and elect its nominee.

More precisely: some voters might not be morons. But that is irrelevant – the same election results arise, as if they all were.

But before you get too enthralled with the Lichtman "13 keys" method, we must warn you that there are three reasons things are less rosy for Lichtman than we just made it sound. First, many of his keys are not completely objective yes/no questions. They depend on evaluations of such subjective words as "serious," "significant," "economic recession," "charismatic," "hero," "failure," "scandal," "major changes," "sustained social unrest," and "success." That leaves Lichtman a lot of "wiggle room" to use to pretend his predictions were "successful." And indeed, Nate Silver attempted in 2011 to redo the Lichtman-keys calculations independently himself, using his own subjective judgments where necessary – with his conclusion being that they actually mispredicted the 1992 election. Silver also disagreed with Lichtman about a couple of other "keys" predictions during 1860-1980, and guesstimates that the method really will be about 80% accurate.

Second, since his system predicted Gore would win in 2000 – oops, actually Bush did – Lichtman then qualified by saying his predictions pertained to the popular vote winner (who was Gore) not the official (electoral-vote-based) winner. The trouble with that is, in 2016, Hillary Clinton was the popular-vote winner, while Trump won only due to the machinations of the electoral college. Oops again.

Finally third, and probably most importantly, let me express a certain amount of skepticism re voodoo. Specifically, here is how to pretend to be a presidential-prediction genius. Begin by constructing a set of 10 binary questions whose answers seem coin-toss-like, i.e. essentially totally random and independent, although actually wholy deterministic.


You get the idea.

Note, unlike Lichtman, my questions are totally objective – I'm not going to (in cowardly fashion) allow any "wiggle room"! Also, I'm not going to weasel about "popular vote"; I'm going to predict the official winner! And all my questions are going to be completely unrelated to history, voter thinking, polls, candidate statements, etc, i.e. my model's voters truly, beyond any shadow of a doubt, are effectively total moron-robots. And to put the icing on the cake, I am not going to need 13 questions. I'm only going to need 5, because I'm way more of a genius than Lichtman – i.e. my questions are way more insightful than his, in some deep mysterious way not at all apparent to anybody reading them.

Theorem: Among 10 such questions, it is likely that there exists a 5-element subset of them, and a labeling of each of their answers as "positive" or "negative," such that the presidency, the last 9 times in a row, was predicted correctly every time, by "were the majority among the 5 question-answers positive?" (And if such a test exists, then one can find it, i.e. find which subset and which labeling, by computer exploration of all possibilities.)

Proof: The number of 5-element question-subsets is 10!/5!2=252. The number of ways to sign-label those 5 questions is 25=32. So, in all, the computer explores 252×32=8064 possible tests. Each possible test has a 2-9=1/512 chance of successfully predicting all 9 binary-choice presidencies in a row. So therefore the expected number of 100%-successful tests is 8064/512=15.75. With almost 16 working tests expected to be found by my computer, it is likely that ≥1 exists. Q.E.D.

Actually, the use of the undefined word "likely" means our "theorem" is not really a theorem. Real theorems must be completely unambiguous. But we think this is good enough to make our point.

(And to make it very likely this ploy will work, instead select the best 7-question subset among 14 such questions, not 5 from 10.)

The moral of this little math-exercise is: Beware statistical humbug! This sort of thing is a standard scam for astounding the gullible with your "genius" and the "immense insight" of your "amazing system." In reality, it may not necessarily tell us anything useful.

I do not know the whole story of the "13 keys" method so I am not willing to claim it necessarily is total garbage. I just recommend caution. Also, if it really is as effective as Lichtman claims, that is a frightening thing and seems to imply that democracy is virtually pointless.

Approval voting (with "primary" stages eliminated by merger so just one combined election)

We want to assess who would have won the presidency if all major primary contenders had run together in a nationwide election conducted with approval (or score) voting. To do that, we collect polls that (1) employed those voting systems, and (2) enquired about these candidates. Such polls of course only happened during the "primaries." Primary voting in different US states, took place from 1 February to 14 June 2016. Therefore, perhaps polls during the middle of this period, i.e. April, seem the most relevant. The following table summarizes approval-style polls sampling over 1000 people each (nationwide adult) taken from February to July. The candidates are listed in descending order of fav/unfav ratio.

CandidateFavorable%Unfavorable% Polls used
Sanders 41 to 6028 to 43CBS, CNN/ORC, ABC/WP, PRRI/A, Bl/S
Kasich 37 to 4924 to 32CNN/ORC, Bl/S
Rubio 37 to 4539 to 45Qu, CNN, ABC/WP
H.Clinton 31 to 4849 to 56CBS/NYT, Bl/S, CNN/ORC, ABC/WP, PRRI/A
Jeb Bush 3748Qu
Trump 24 to 4352 to 68CBS/NYT, Bl/S, CNN/ORC, ABC/WP
Cruz 13 to 3648 to 57CBS/NYT, Bl/S, CNN/ORC, ABC/WP, PRRI/A
CBS       CBS News Poll, ≈1131 reg.voters USwide, 7/29-31, 5/13-17, 4/8-12
CBS/NYT   ≈1333 reg.voters USwide, 7/29-31, 7/22-24, 7/8-12, 6/9-13, 5/13-17, 4/8-12, 3/17-20
CNN/ORC   1016 adults USwide, 7/29-31, 7/22-24, 7/13-16, 6/16-19, 4/28-5/1, 3/17-20, 2/24-27
ABC/WP    ≈1100 adults USwide, 9/19-22, 8/24-28, 8/1-4, 7/11-14, 6/8-12, 5/16-19, 4/6-10, 3/3-6
PRRI/A    Public Religion Research Institute/The Atlantic. 3/30-4/3. 2033 adults USwide.
Bl/S      Bloomberg/Seltzer 1000 adults USwide 8/5-8, 6/10-13, 3/19-22
Qu        Quinnipiac Univ. 1342 registered voters USwide, 2/10-15

As you can see, Sanders would have won the election if approval voting had been used and the election had been held during the primaries period.

What about Kasich, who came the closest to Sanders in approval? In the 12 pairwise Sanders-vs-Kasich polls compiled by during March, April, and May 2016, Sanders won 9 and Kasich 3 of them. Kasich's largest margin of victory was 44-41 and Sanders' largest 52-41. Also Sanders won the final 7 among these 12. So I think Sanders' chances would have been about 80% in a head-to-head race versus Kasich.

It is perhaps worth reiterating that both Clinton & Trump were the least-approved major party US presidential nominees in the history of approval-style polling. This is not a close call.

The absurdly anti-democratic nature of US primaries. It also is worth noting that Trump got 14.0M votes in the Republican primaries, and Clinton 16.85M in the Democratic primaries, i.e. together 30.9 million votes. This is 9.6% of the USA's estimated 319.9M population, and an extremely unrepresentative 9.6% too.

Score voting (with "primary" stages eliminated by merger so just one combined election)

NBC News/Wall Street Journal/Hart Research poll (10-14 April 2016, 1000 Registered Voters, phone) "Now I'm going to read you the names of several public figures, groups or organizations, and I'd like you to rate your feelings toward each one as very positive (4), somewhat positive (3), neutral (2), somewhat negative (1), or very negative (0). If you don't know the name, please just say so (X indicates don't know or refuse).":

Candidate01234 Xmean
J.Kasich 8%1131238192.27
B.Sanders 221416242132.08
"Ted" Cruz30192117941.54
H.Clinton 421412191301.47
D.Trump 531210131111.16
Paul Ryan*111728179181.95

(Notes: The numeric values 43210X were not mentioned to the pollees but are used for our average-computing purposes. Obama and Ryan were not actually presidential candidates in 2016 and were included purely for comparison purposes. The actual poll wording stated both names, e.g. "Donald Trump" not "D.Trump." These same notes apply to all the score-style polls in this section.)

Investor's Business Daily/TIPP telephone Poll, TechnoMetrica Market Intelligence, 537 cellphone, 306 landline (random digit dial). 22-29 April 2016. 903 Americans aged≥18. "Please tell me, generally speaking, if your opinion of the candidate is Very Favorable (3), Somewhat Favorable (2), Somewhat Unfavorable (1), or Very Unfavorable (0)." X indicates don't know/refused:

Candidate0123 Xmean
B.Sanders 230116242183431.49
J.Kasich 11217722878101.46
H.Clinton 321135198148731.22
"Ted" Cruz310160201702191.04
D.Trump 41292162131171.02

PRRI/TheAtlantic April 2016 Survey. Total=2033 (813 Landline, 1220 Cellphone). 30 March to 3 April, 2016. "Now we'd like your views on some political leaders. Would you say your overall opinion of [INSERT;RANDOMIZELIST] is very favorable(3), mostly favorable(2), mostly unfavorable(1), or very unfavorable(0)?" X indicates don't know/refused:

Candidate0123 Xmean
B.Sanders 20212918121.51
H.Clinton 3420231761.24
"Ted" Cruz2928227141.08
D.Trump 521816860.79

Note: Kasich was not included in the PRRI/Atlantic poll. PRRI=Public Religion Research Institute.

Approval voting (official post-primary contenders only)

The poll-aggregators and Huffpost Pollster helpfully compiled favorable/unfavorable ratings (i.e approval-style polling; our RCP poll-averages covered the period 28 Oct to 7 Nov 2016 while HuffPost via "trendlines" for their poll-collection attempted to produce results valid for election day 8 Nov). The table gives their final pre-election-day results.

CandidateRCP fav/unfavHuffPost fav/unfavSUUT fav/unfav
Gary Johnson 20.8/39.7=0.524154/342=0.45
Jill Stein 17.4/37.0=0.470125/258=0.48

For those who want a particular poll, not a (dubious?) aggregration, one was conducted by Suffolk University for USA Today and is found in the rightmost column of the table above. It is the chronologically last nationwide poll I am aware of which asked an approval-style question for all 4 nominees. [1000 likely voters ages≥18, live telephone interviews, 20-24 October 2016, residing in all 50 states and the District of Columbia, who said they intended to vote in the general election on 8 Nov.]

Johnson's F/U ratio 0.524 (claimed by HuffPost) is confirmed by a poll we spoke previously about, which was not part of the HuffPost aggregation. It was conducted by the Johnson/Weld campaign on 8-9 October, and found ratio 0.507.

As you can see, Clinton would have won with approval voting (nationwide popular vote) with this slate of 4 candidates on election day. (She still would lead even if we subtracted 5 points from each of her "favorable" percentages.)

The main reasons Clinton's lead widens with approval versus the official plurality voting margin, presumably are

  1. Because of the elimination or reduction of the "spoiler effects" from Johnson and Stein.
  2. Because Trump-disapprovers were less likely to also-disapprove Clinton, than Clinton-disapprovers to also-disapprove Trump.

It also is worth noting that Hillary Clinton's low approval ratings are of recent vintage, as we see in this graphic from Gallup:

In contrast, Trump actually appears to have gradually become less disapproved over the 1.5 years preceding the election, moving from about 23/62 Fav/Unfav rating in early July 2015, to about 37/58 in late Nov. 2016.

Note also that every candidate, unfortunately, was more disapproved than approved. But oddly enough the outgoing president Barack Obama (forbidden by the 22nd amendment from seeking a 3rd consecutive term) enjoyed majority approval at this same time, i.e. would have won re-election by landslide (if that were legal and he'd been added to this slate of candidates; see bottom row of table, shaded blue). Indeed, at the end of his second term, Obama enjoyed 55/41 fav/unfav rating.

Also note, that Clinton's approval lead that we have assessed was on election day 8 Nov. In contrast, in mid-September, the same methodologies concluded that Johnson led approval-style polling.

Score voting (official post-primary contenders only, but sometimes a few wildcards added)

Here are the results of a Franklin Pierce University/Boston Herald nationwide poll conducted by RKM Research, 31 August to 4 September, 2016. The survey is based on responses from 1025 randomly selected likely voters via interviews conducted by landline and cellular telephone. The scale is VU=very unfavorable=0, SU=somewhat unfavorable=1, SF=somewhat favorable=2, VF=somewhat favorable=3, with the numbers {0,1,2,3} not told to the pollees but used by us to compute average scores. The table entries say the percentage of respondents who chose each score for the candidate in that row. (Row sums less than 100 because of "don't know" responses, not shown.) Besides the 4 candidates, they also included in their poll Bernie Sanders (whose run ended after his defeat by Clinton in the Democratic party primary), Elizabeth Warren (who never ran, but had often been mentioned as a possible Dem-party contender) and Bill Weld (who also did not run for President but was Johnson's vice presidential running mate). Their table lines are shaded green and blue.

Candidate VUSUSFVFaverage
Bernie Sanders(D*)241426281.63
Elizabeth Warren(Dx)221017251.61
Gary Johnson(L)391041.58
Bill Weld(Lx)25911.53
Jill Stein(G)36721.44
Hillary R. Clinton(D)44921211.20
Donald J. Trump(R)50916211.08

As you can see, if these candidates had run using score voting at that time (slightly over 2 months before election day), then Sanders would have won, with Trump finishing in last place; and Johnson would have been the winner among those actually still running at the time.

Independent confirmation of Johnson's victory (and of the ordering of everybody besides Stein) is provided by a nationwide poll by Quinnipiac University on 8-13 September 2016, among 960 likely voters using interviews on both land lines and cell phones and speaking in either English or Spanish dependent on respondent preference. Here are the Quinnipiac results:

Candidate VUSUSFVFaverage
Gary Johnson(L)15121461.23
Hillary R. Clinton(D)51614261.15
Donald J. Trump(R)53613221.04
Jill Stein(G)135530.92

But next is a poll by Selzer & Co. for Bloomberg News, conducted 21-24 September among 1326 general population contacts ages≥18 (landline and cellphone interviews) of whom only the 1002 "likeliest voters" were used. Note that now, things have changed; Trump rose and Johnson sunk. However, all three of {Trump, Clinton, Johnson} really are tied in the sense that their differences lie within the statistical margin of error.

Candidate VUSUSFVFaverage
Barack Obama(D*)341320311.49
Paul Ryan(R*)172832111.42
Donald J. Trump(R)431221221.22
Hillary R. Clinton(D)411524181.19
Gary Johnson(L)17232131.16
Jill Stein(G)19191331.00

You may object that polls held in September were too early. Fortunately for you, the FPU/Boston Herald/RKM nationwide poll was redone on 1-5 November (1009 random likely voters nationwide; landline & cellphone), and it turns out the "too early" criticism indeed carries weight, because during the two months between these two FPU/BH/RKM polls, things changed! Most prominently, Johnson's unfavorables rose. Here are this poll's results:

Candidate VUSUSFVFaverage
Donald J. Trump(R)
Hillary R. Clinton(D)47.48.717.524.91.20
Jill Stein(G)12.915.
Gary Johnson(L)18.118.714.41.81.00

Incredibly, the ordering-by-average-score of the 4 candidates has reversed versus the earlier Aug-Sept poll by the same pollster using the same methodology! (And although Trump "won" this poll, actually both he and Clinton are co-winners in the sense that their difference lies within its statistical margin of error. Also note, this poll's ordering is confirmed, except for Stein, by the preceding Bloomberg/Selzer poll.)

Finally, a post-election poll (10-14 Nov. 2016, interviewed 1254 registered voters aged over 18 who said they had voted, telephone both landline and cell) by Pew Research Institute asked voters to "grade" the candidates A,B,C,D,F. Only 30% gave Trump either an A or B, the lowest percentage among winning candidates in every US presidential election 1988-2016, and also lowest – but tied with G.H.W.Bush in 1992 – among second-place candidates in those years. (Note, 43% gave Clinton an A or B, and a higher average, marking the only time 1988-2016 that the loser actually outperformed the winner on Pew's grading question.)

(Percentages awarding each grade shown.)
Who ABCDF Average grade
Hillary Clinton 1231201621 C
Donald Trump 921191535 C-
Republican Party 914252230 D+
Democratic Party 719262028 C-
The Press 616192138 D+
The Pollsters 615242130 D+
The Voters 1822291513 C+

Suicide-level idiocy of both the Republican and Democratic (and Green!) parties

During the primaries, there were many "pairwise polls" attempting to assess who would win, X or Y, in a head-to-head X-versus-Y race. (For almost all of them, X was a Democrat and Y a Republican.) These polls indicated that in a Clinton-versus-Y race, or Sanders-versus-Y race, Trump was one of the worst possible choices for the Republican Y, i.e. one of the least likely to win that pairwise battle. They also indicated that in a X-versus-Trump race, Sanders clearly outperformed Clinton. There also was a round-robin poll by GfK for the Center for Election Science (sample>1000) indicating Sanders actually would have defeated every one of 8 rivals pairwise. (Unfortunately the 8 "rivals" they assessed did not include John Kasich, but did include Michael Bloomberg, who decided not to run. But when we found 12 other pairwise Sanders-Kasich polls ourselves they suggested Sanders would have won that pairing too):

honest opinions pairwise table by CES 2016

So the Democrats were suicidally stupid to nominate Clinton, who lost, as opposed to Sanders, who would have won. Meaning they were suicidally stupid to

  1. Employ plurality voting;
  2. Usually demand only registered Democrats vote in their primary, refusing to permit either Independents and Other-Party members to vote;
  3. Even in "open primary" states, or states where Independents were legally permitted to vote in the Democratic primary, the Democratic party (in at least some cases) corruptly discouraged this and tried to keep it quiet.

Because: if the Democrats had instead employed approval or score voting for their primaries, and permitted Independents to vote in them (or merely the latter alone), then Sanders would have won their nomination, and then the Presidency. And further, this also would cause the Democrats to nominate candidates more likely to win the presidency, every time, not just in 2016.

Also: The Democratic nomination hinged on the "superdelegates," who were non-democratically chosen. Almost all superdelegates chose Clinton. If, instead, almost all had chosen Sanders, then it is a mathematical fact that he would have been nominated, and then a polls-and-election-results-based fact that he thereupon would have won the presidency. The latter is essentially certain based on the gap between Clinton and Sanders in X-versus-Trump pairwise polls, versus the small official Trump-Clinton margins in the three crucial states of MI, WI, and PA.

The Sanders team had in fact pointed out the X-versus-Trump pairwise poll evidence in an unsuccessful attempt to get the superdelegates to switch. And Sanders was entirely correct in this argument. The reason he failed to convince the superdelegates of this convincing and correct argument was presumably that the superdelegate system is massively corrupt – based on political favors, money transfers, and influence peddling, and not on poll-based realities about what is best for the Party (which is the alleged raison d'etre for the superdelegate system). Clinton had had 20 years as a top Democrat to suck up to superdelegates. But Sanders had only become a Democrat (previously Independent) for his 2016 presidential run, and therefore could not compete in this sleaze-battle.

Sanders vs TrumpIn the 21 pairwise polls compiled by during March, April, and May 2016, Sanders won them all. His smallest margin of victory was 46-42 and his largest 58-34.
Clinton vs TrumpIn the 59 pairwise polls compiled by during March, April, May, and June 2016, Clinton won 52 and Trump 6 of them, with 1 tie. Trump's largest margin of victory was 42-37 and Clinton's largest 54-36. [These results would seem, a priori, to give Trump about 11% chance to win this pairing.]

See also [Ryan Rifai: Polls: Sanders has more potential to beat Trump, Al Jazeera 14 May 2016], which noted "Sanders [averaged] a 13 percent advantage over Trump, while Clinton had 5." It also said

Dustin Woodard, an analytics expert who played a major part in the discovery of the Reuters poll trend, told Al Jazeera that a significant reason for Sanders' advantage was due to disproportional support from independent voters... Independents are the largest voting population in the US. Gallup reports that independents are 42 percent of the voting population, while Democrats are only 29 percent and Republicans 26."

See also Dan Hopkins: Why Sanders Does Better With Independents, 18 April 2016,, which noted

In New Hampshire, for instance, Sanders won Democrats by 4 percentage points while winning independents by nearly 50... in Ohio, Sanders won 66 percent of independents but just 35 percent of Democrats... [this is] a split we've seen repeatedly since then.

Similarly [Ryan Brownstein: A Primary That Pitted Democrats Against Independents, The Atlantic magazine 5 June 2016]

Results from the exit polls conducted in 27 states through the nominating contest so far show that Hillary Clinton has established a huge lead over Sanders among voters who self-identify as Democrats. Sanders, though trailing in the popular vote and delegate count, has remained competitive only because he has built a virtually identical lead among primary voters who self-identify as independents. This pattern has persisted across all regions of the country...

For yet more evidence, see the PRRI/Atlantic poll released 7 April.

Support among both Democrats and Independents (and everybody else) is what matters both for benefitting the country, and also for the chances of winning the general election. Our point is that Sanders was preferred within this combined set because Independents outnumbered Registered Democrats; and the Democratic party was, in this election, suicidally stupid to try to discourage/stop Independents from voting in their primaries.

Meanwhile, the Republicans were suicidally stupid to nominate Trump, who had among the worst chances versus Democrat X, as opposed to almost anybody else, who had better chances. Trump won that nomination solely due to the fame-based failure pathology of plurality voting, which made the Republicans look like utter jackasses once again. They would have avoided this problem with approval or score voting. Then Trump would not have won their nomination. And this would give the Republicans better chances to win the presidency not just in 2016, but every time.

Clinton vs KasichIn the 15 pairwise polls compiled by during March, April, and May 2016 (were none in June), Kasich won them all. His smallest margin of victory was 45-41 and his largest 51-39.
Clinton vs TrumpIn the 59 pairwise polls compiled by during March, April, May, and June 2016, Clinton won 52 and Trump 6 of them, plus 1 tie. Trump's largest margin of victory was 42-37 and Clinton's largest 54-36.
Sanders-Trump and Kasich-Clinton pairwise polls USA 2016

Incidentally, there are also were pairwise polls concerning only Republican primary candidates, asked only of Republican primary voters:

Cruz:Trump NBC/WSJ(Feb): 56:40. NBC/WSJ(Mar): 57:40. ABC/WP(Mar): 54:41.
Rubio:Trump NBC/WSJ(Feb): 57:41. NBC/WSJ(Mar): 56:43. ABC/WP(Mar): 51:45.
Kasich:Trump NBC/WSJ(Feb): 44:52. NBC/WSJ(Mar): 57:40.
NBC/WSJ: NBC News/Wall Street Journal Poll conducted by Hart Research Associates and Public
Opinion Strategies. Dates 14-16 Feb 2016 (400 sampled), and 3-6 March (397 sampled).
ABC/WP: ABC News/Washington Post Poll 3-6 March 2016, asked of 
400 registered voters nationwide who are Republicans or lean Republican:
"What if the choice was just between Trump and Cruz? Who would you like to 
see win?" If unsure: "Who do you lean toward?" 
The two numbers sum to less than 100, such as 57+40=97, because of "don't know" voters.

These polls make it clear that Trump indeed won the Republican nomination solely as the result of the "fame-based failure" pathology of plurality voting – i.e, his rivals split the vote. If pitted against any of these three rivals head-to-head, Trump would have lost. (The only poll disagreeing with this conclusion is the italicized Kasich:Trump poll in February, won by Trump; but it is outweighed by the March re-poll with larger and opposed conclusion.)

In case you were wondering, the Democratic primary also was severely distorted by its plurality-style voting. Consider Martin O'Malley, the only life-long Democrat in the Democratic field (and past chair of the "Democratic Leadership Council"). He was mayor of Baltimore, two-term governor of Maryland, and notched several progressive achievements, including passing legislation legalizing same-sex marriage and outlawing the death penalty. Why was all that only worth about 1% in national polls, causing O'Malley to drop out of the Democratic race in frustration after getting 1% in the Iowa Caucuses on 1 Feb? Obviously, this again was a huge distortion caused purely by "must vote for one of the top 2 leaders" strategic imperatives with plurality voting (the "top two most-likely-to-win" rapidly coalesced as being Bernie Sanders and Hillary Clinton) and had little to do with O'Malley's true level of support.

Finally, the Green party actually came out with videos supporting instant runoff voting as the cure for the USA's ills. That was yet another suicidally stupid move.

Why? Because in this particular election there is no question whatever that IRV would not have altered the election result. And more generally, IRV would not get rid of the USA's 2-party domination that has prevented the Green party from ever winning even a single federal seat ever. The Australian Green party actually did have one of its members win a federal House seat in Australia via a standard IRV election (Australia elects its House with IRV). I repeat, one. His name is Adam Bandt. He was the only third-party member ever to win an Australian House seat in a standard IRV election during 1950-2015. This perhaps is why the Australian Green Party calls for the abolition of IRV for electing its House. And why the Australian public as a whole (in polls) keeps saying they would abolish IRV for electing its House, if given the chance in a referendum.

The USA Greens, were they not this awe-inspiringly stupid, would advocate score or approval voting.


These smug pilots have lost touch with ordinary passengers like us. Who thinks I should fly the plane? – Caption of New Yorker cartoon showing an outraged passenger on an airplane drumming up support.

Plurality voting: Clinton won the nationwide popular vote by about 2% over Trump. But Trump won the presidency thanks to the "electoral college."

In the below considerations about non-plurality voting systems, we will speak only of popular vote – except that in the case of IRV what we say is true regardless of whether electoral or popular is used.

Score voting (restricted to the four final party-nominees Clinton, Trump, Johnson, and Stein only): The outcome is very unclear. All three – Clinton, Trump and Johnson – led, depending what time the poll was taken during the final 2 months (plus the 1 week immediately after election day) and perhaps Stein also could have won. However, it appears any one of

Bernie Sanders, Elizabeth Warren, Barack Obama, Paul Ryan

(listed in approximately descending order) would have (if they too had run) easily defeated all four of these.

Approval voting (restricted to those four): Hillary Clinton wins. But again if Obama had been allowed to run (or if Sanders or Kasich had continued to run), either would again have easily defeated all four. Also Johnson led back in mid-September, as opposed to on election day.

Instant Runoff voting (restricted to those four): The electoral and popular results would have been the same as with the official plurality-based system(s).

Incidentally, the CES/GfK poll found that instant runoff exhibited a 7.7% "favorite betrayal" rate, i.e. 7.7% of their sample of over 1000 (polled about a 9-candidate field) said that they, if using instant runoff voting, would have chosen to dishonestly rank their true favorite below top. This actually was a greater favorite-betrayal rate than the 7.3% that this same poll found for plain-plurality voting, contradicting common claims by IRV-proponents (those claims unfortunately are usually made with zero supporting evidence) that IRV would cause greater voter honesty and less strategic lying than plain plurality voting.

Approval Voting (all primary contenders, all parties): Sanders would have won, with Kasich second. Oddly(?) enough, these also were the two candidates with the greatest truthfulness percentages according to fact-checking agencies; and also Kasich would seem to have been the most-qualified Republican.

Score Voting (all primary contenders, all parties): Either Sanders or Kasich would have won.

Instant Runoff Voting (all primary contenders, all parties): It would have been utterly absurd to try to get voters to rank-order about 25 candidates. For this reason, no reputable pollster conducted an IRV-style poll (and probably none ever will in any comparable circumstance). There were, however, a few "second choice" polls and based on them we can try, somewhat dubiously, to estimate the IRV winner; e.g. probably the Republican nominee with IRV would have been either Walker or Trump, and the Democrat still would have been Clinton.

Suicidal Idiocy Tally: Both the Democratic and Republican parties were foolish to use the voting systems they did, since each would have assured winning the presidency by switching their primary to use approval voting with Independents allowed to vote too. The Green party foolishly advocated Instant Runoff Voting which would not have altered the winner and would not help US third parties; but with score voting Gary Johnson actually might have won the 2016 presidency.

Partial Damage Tally: Trump was (simultaneously) the uniquely least qualified, lyingest, most financially interest-conflicted, oldest, most insanely litigious, most divorced, least-approved, most newspaper-opposed, and most-massively-bankrupt US president in the history of record keeping for each of these categories. He promised he'd militarily force one of the largest migrations in human history. And over 20 women publicly accused him of sexual assault.

Return to main page