Why we should adopt multi-round preferential voting

18 Jan

It’s very commonly suggested that the American political process is unfairly captured by the two major parties and that we should rectify this by adopting instant-runoff voting, a.k.a. preferential voting. In such voting schemes, instead of simply choosing a single most preferred candidate, a voter may rank all candidates in order of preference, with the idea that this makes third-party candidates more viable. For example, if I most prefer the Libertarian party candidate but also prefer the Republican candidate over the Democratic candidate, then I can express these preferences without concern that my vote won’t advance the Republican over the Democrat.

The important questions about an instant-runoff voting system are:

  1. Would it actually make third parties viable?
  2. Would that be a good thing?

My answer on both counts is, ‘probably not’:

Assume for a moment that we can accurately distill voter preferences into a single dimension of ideology, such that a given electorate has a collective preference for a candidate on a particular spot along the left-right, liberal-conservative axis. What we tend to get in our elections, such as our US House and Senate races, is a dynamic wherein the Democratic and Republican candidates occupy the middle of both sides while 3rd party candidates flank them on the ends. Though many members of the electorate identify as centrists, in truth they mostly sit ideologically clustered into two major peaks not much further towards the center than the Democratic and Republican candidates.

So assuming that the ideological positions of the two major parties, the third parties, and the electorate remain the same after the introduction of preferential voting, what then would be the point? Democratic and Republican candidates already best represent the electorate’s ideological preferences, so third party candidates would never win without shifting to the same positions, and if third parties can only win by becoming just like the two major parties, why do we need third parties? Sure, unpopularity isn’t the only reason that third parties can’t compete in our current system, but it’s likely the primary reason: given their current policies and ideology, the positions of the Green Party, the Libertarian Party, the Constitutional Party, et. al. simply aren’t popular, and preferential voting won’t change that.

Which is not to say that American democracy doesn’t have a systemic gap between what the electorate wants and what it actually gets. I just think that proponents of third parties have the wrong diagnosis and hence have the wrong prescription.

I actually agree that we want a preferential voting system, but one modified in a few crucial ways. The preferential voting system we need follows a simple strategy, the same strategy which people commonly use to make optimal choices in other contexts. This strategy starts by, from the set of all possibilities, identifying a subset of the most favored choices, then in successive rounds, whittling that subset into smaller subsets until left with a single choice. So for example, in a 3-round system, the voters might collectively select the top 30 candidates in round 1, then, from that 30, select the top 6 in round 2, and then, from that 6, select the final winner in round 3. The last round would use Instant Runoff Voting, and all previous rounds would use Single Transferable Vote, a preferential voting system that selects multiple winners.

Now, in a sense, our current system is already a multi-round system because we hold primaries to select party nominees. The scheme I have in mind, however, gives no official role for political parties. If a party wants to select just one member to appear on the first-round ballot, that’s the party’s business; its members would all still be free to appear on the ballot under their party affiliation. So even if, say, the Democratic Party selects one nominee, any other candidate is free to appear on the first-round ballot as a Democrat. If a party wishes it to be known which single candidate it actually endorses, that’s the party’s responsibility. Any candidate could still appear on the first-round ballot claiming to affiliate with any party.

So consider, again, how our current elections play out:

The electorate in the US is clearly ideologically clustered around two peaks, such that only the Democrat and Republican nominees have real chances of winning in the vast majority of our elections. The third parties simply can’t compete because they occupy the wrong ideological territory. Because most voters have a partisan Democratic or Republican affiliation, they end up voting for their party nominee, even if they don’t like the nominee very much.

A non-partisan, multi-round preferential voting system, however, should work quite differently. Assuming again a 3-round system, the first round would whittle down the space of all possibilities down to a more manageable number, say 25 or 30, and then the second round would whittle the field down further to say, 5 or 6. Because each round selects the most preferred candidates of the entire electorate rather than just one party, the final ballot will likely then consist of several very competitive candidates, even in districts usually dominated by one party. In a heavily Republican congressional district, for example, a final ballot of 5 candidates might include 4 Republicans and 1 Democrat. Thanks to instant-runoff voting, the Republican voters would rank their preferences without concern of throwing the election to the Democrat.

What, though, would be the point? In all likelihood, this electoral process would produce Republicans and Democrats who occupy more or less the same positions on the ideological spectrum as they do today. If Republican districts still elect Republicans, Democratic districts still elect Democrats, and competitive districts still swing back and forth, what then would be the advantage?

Well, consider three problem cases in our current voting system:

  • First, in theory, an incumbent with an unpopular position should get voted out, but partisan voters are very reluctant to contest their own incumbents, mainly because primary challenges significantly increase the odds of handing the seat to the opposition.
  • Second, underfunded candidates with low name recognition are greatly disadvantaged. Even when such candidates best matches voter preferences, they tend to get little attention from funders and voters because they lack legitimacy, a sense that they could actually win.
  • Lastly, consider a candidate who deviates from their party on one or two major issues. Even if such a candidate most closely mirrors the electorate, they generally get filtered out by the party primary, where candidates must appeal first to the party base.

In all three of these cases, our current voting system often filters out candidates who best match voter preferences. Multi-round preferential voting shouldn’t suffer the same problems:

  • With preferential voting, voters need not worry about throwing their votes away and so can express their true preferences. Thus, incumbents who deviate from the electoral will on one issue would be more likely to get replaced by a similar candidate without that deviation.
  • Lesser-known candidates would have better odds of winning because making it through the first round of voting should itself lend them some credibility.
  • Because candidates needn’t pass through a party primary, they would be able to mix-and-match ideological positions on different issues.

So, that’s the gist of why I favor multi-round preferential voting. Of course, a new voting process isn’t a panacea for fixing our politics. Our deeper problems stem from the dozens of veto points in our legislative process that severely mute the consequences of our elections. I have some ideas on how to reform our legislative process, but I’ll save them for a later post.

Android should get rid of the back button.

13 Nov

I made the longer argument here, but let me make it more succinctly. The chain of logic goes:

  1. The back button should only be used for navigating between apps, not within any application.
  2. The recent apps screen serves application switching more predictably than the back button (even under its most favorable circumstances), except…
  3. …the recent apps menu is not quick enough for the common case of switching back to the last app, so we need a quick button/gesture for this. I suggest a swipe up motion from the recent apps button (similar to the swipe up on the home button that currently takes you to Google Now). It’s debatable whether repeating this gesture should cycle you through all recent apps or just between the current and last app (I suspect current and last is the better behavior). Either way, this back-to-last-app allowance covers the most useful case for the back button: when an application switches you to another app–such as to display a link in the browser, compose an email in the mail app, or play audio in the music player–you should be able to get back to what you were doing previously with little effort or thought.

So the recent apps button with a simple new allowance is a better back button than the back button itself.

Now of course, claim (A) requires substantiation, which two members of the Android team do a well enough job providing here. The odd part is that they as much admit in passing that the back button is a bad in-app navigation mechanism and then spend the rest of the talk proudly describing the surprisingly convoluted scheme Android and its apps must conform to to deliver non-maddening  back button behavior. Why subject developers and users to such headaches?

Android should ditch its too-clever-by-half task/activity paradigm in favor of the app/screen model expected by developers and users alike. If my app wants to open a link in the browser, that’s not a new activity, its just an app switch to the browser, just like on the desktop; going back to the app should be an ordinary matter of app switching requiring no special cooperation on any app’s part.

A problem that remains is deciding when apps should open to their last viewed screen or their ‘root’ screen: when I switch back to the app I was using 2 minutes ago, I usually expect the screen I was just looking at; but when I switch back to an app I haven’t used in a long while, I usually expect the app’s root screen. I’m not sure what exactly the timeout period should be, but I do know what screen I see shouldn’t depend upon how I open or switch to an app, whether through the launcher or recents list. It should be the responsibility of apps, not Android, to make their root screens easily accessible from any state.

Again, strangely, the Android team seems to have caught on to this, at least going by the example they’re setting in their own apps with the ‘Up’ navigation element now in the Gmail, Caldendar, and Youtube apps. Obviously some legacy apps rely upon the back button, so it should get deprecated like the menu button: back should only contextually appear on the bottom bar within those legacy apps. And of course, some applications might have their own in-app need for history navigation, e.g. the web browser. This in-app navigation, though, should be done with app-specific, in-app controls, not a universal button. Again, the Android team already sets the precedent here by including a proper back button in Chrome for Android.

The last remaining piece is to make app-initiated app switches visually distinct from mere in-app screen switches so as to keep the user oriented (and so the user can utilize the back-to-last-app action when it’s most needed). If an action in an app takes me to the web browser, Android should make it very clear that I’m switching to another app. One solution would be to replace the ‘recent app’ button icon with the icon of the current app, such that the user can always get a visual indicator of the current app; when switching apps, this icon would pulse or glow for a brief time to make the switch more noticeable.

Everyone Is Awful

9 May

I’m not the person to adjudicate whether our economic problems are “cyclical” or structural. But the reason I find Krugman’s anti-structural take persuasive (that is, aside from his arguments and facts) is that I think I understand why so many people want to see our problems as structural. It’s part of a broader notion I call Everyone Is Awful. Briefly, Everyone Is Awful is the belief that people today are worse than ever–stupider, lazier, less moral–and that this worsening of our culture and human stock underlies all of our largest problems. Basically, it’s the Kids These Days trope on steroids, elevated to a grand, world-historical narrative.

Applied to economics, Everyone Is Awful-thinking concludes that, ‘of course we have mass employment because a large chunk of us are less employable than ever’. Not only is this narrative factually wrong, the growing reach of the Everyone Is Awful narrative is itself partly to blame for the very ailments it purports to explain. If we have a real long-festering structural problem in our economy, it’s that elites have become allergic to risk; more than ever, the investor and management classes want a sure thing. This gives us financial markets that disastrously fool themselves into thinking that they’ve engineered away risk (or sloughed it off to others) and a hiring environment that expects job seekers to come fully formed as totally known quantities, predictable cogs that fit a precise purpose and which can be disposed at the first sign of deviation from specification. If an investment isn’t a perpetual money machine, it’s too uncertain; if this potential hire doesn’t have a prestigious degree and previous success with this precise area of practice, we can’t take the risk. Only star performers have real value, for whatever value non-star assets and people might have, that value is too hard to discern and so must be discounted. In short, everything has degraded to a lemon market.

How do we combat this? I have no idea. But notice that in the Everyone Is Awful narrative, elites are faultless, having been only failed by the common people. Should we be surprised that they like structural stories?

*To be clear, the ‘Everyone’ in Everyone Is Awful means ‘other people’.


Not All Bad Laws are Unconstitutional

29 Mar

Matt Yglesias and several of his commentators observe that nothing in constitutional law demands a limiting principle to the commerce clause or, in fact, to any enumerated power. Perhaps the framers didn’t intend Congress to impose its will on any domain of life through taxation and regulation of commerce, but if that was the founders’ intention, they simply fucked up in writing the Constitution. (Who said the Constitution can’t possibly contain mistakes?) The commerce clause allows regulation of international and interstate commerce; if it’s found that nothing in modern life stands outside of interstate commerce anymore, then nothing stands outside Congress’s enumerated powers. That the founders didn’t foresee this evolution in the 1780′s doesn’t obligate us to invent new limitations.

More broadly, Matt and other commentators note that plenty of bad, liberty-infringing laws are perfectly constitutional. There’s no special obligation to limit the commerce clause to preclude such laws, and very importantly, this is no big deal. The primary protections of liberty afforded by the Constitution are in its explicit prohibitions, such as in the Bill of Rights, and in the political process it guarantees. As the McCulloch v Maryland ruling noted, the general protection against tyranny afforded by the Constitution is the electoral process. You don’t want Congress to force you to eat broccoli? Don’t vote for that Congress.

Reverse Godwin’s Law

2 Mar

Regarding Andrew Breitbart’s death

No, not every bad person is equatable to Hitler, but below (above?) Hitler, there are plenty of gradations of bad that are really, really bad. Hallmark doesn’t make a card that reads, “Congratulations, you’re not a Nazi!”

I found the gloating over bin Laden’s death distasteful, so something would be pretty fucked up with my moral compass if I didn’t feel more reservation over Breitbart’s death. By my own sense, the properly calibrated liberal response is a non-gleeful, sober assessment of Breitbart’s life mission—which won’t be terribly flattering, even sans hyperbole.

Presidential primary system reform is doable

1 Mar

Our presidential primary process is:

  • overly complicated
  • overly long

…which are bad things for all the reasons I laid out here. Fortunately, control of the process lies in the hands of the individual parties rather than the federal constitution or federal or state law, so reform shouldn’t require any grand political bargaining across the aisle. Moreover, each party has a strong interest in optimizing the process to select their most viable candidate. Given these facts, why then has the status quo prevailed? The simplest reason is probably that would-be primary reformers simply haven’t settled upon a viable alternative. The most common proposal is a simple “national primary”, which would have a pretty obvious flaw:

The major flaw in the concept is that it takes the phenomenon of frontloading, which other reform plans seek to alleviate, to its ultimate conclusion. Candidates would need to raise huge sums of money, before the first vote was cast in any state, in order to wage a nation-wide campaign. Neither the Republican National Committee’s 2000 Advisory Commission on the Presidential Nominating Process nor the Democratic National Committee’s 2005 Commission on Presidential Nomination Timing and Scheduling considered a national primary as a reform concept; rather they considered it the consequence of inadequate action to reform the process.

The real fault in the idea is that it simply goes a step too far. Yes, we need a national primary, but one with multiple rounds. Adapting my election ideas from the earlier post, my primary system would involve at least two rounds, the first a national STV vote held about four months before the convention to narrow the field down to ~6 candidates, the second a national IRV vote to select the nominee one month before the convention. You might possibly add another STV round in between, selecting ~20 candidates in the first round, ~6 in the second, and a nominee in the final round. Either way, you’d avoid the inevitable front-runner bias of a single-round national primary and probably make the field actually more open than it is currently rather than less. Best of all, you’d greatly simplify and expedite the process, sparing us from the endless 18+ month slog we know today (including the endless media speculation about brokered conventions and county-by-county voting patterns).

If you want to preserve out-sized influence for the smaller states, simply give them a bonus modifier, e.g. one Wyoming vote counts for Nx one Californian vote.

The only real downside I can see is the added cost of holding multiple voting rounds in every state, but it seems like a pretty small price to pay. If this is really an issue, the parties could hold the earlier rounds as private elections, perhaps online (involving, of course, some kind of registered party member authentication and perhaps supplemented with traditional onsite voting). After all, the nomination process is not really a state function, so the nomination contests needn’t really be conducted under the auspices of the state.

If Democrats and Republicans want to avoid a future primary debacle like the 2012 Republican primary, they should give primary reform serious attention.

Advertising is not speech

26 Feb

Matt Yglesias is skeptical that solving campaign financing will cure our politics or even much alleviate the corrupting influence of money. I share this skepticism, and my own take is that we should instead focus on the specific problem of advertising. After all, political donations buy all sorts of things other than television air time or web banners—mainly, organization staff and media production—and I’m not so sure these activities shouldn’t fall under First Amendment protection, even if the rich have a disproportionate ability to partake in such “speech”. More importantly, I think staffing and production are beside the point, as the actual ad buys are where the real money goes. I’d go so far as to say political ad buys are entirely illegitimate: rather than curtailing political ad buys, we should outlaw them outright. The crux is finding a workable definition of “advertisement” that distinguishes ads from non-advertisement speech and then also distinguishes the political from the non-political. This is tricky but I think well doable were it not for the inevitable Supreme Court rollback. I won’t go into the weeds here but just say that any defense of advertisements as a valid form of communication in democratic discourse is laughable.

Speed Racer is one of the best films of the last 10 years

22 Feb

Certainly one of the very best blockbusters. And I’m not the only one who thinks so, apparently. My only major criticism is that it’s maybe 20 minutes too long, but I can’t say specifically what I would cut.

Wisconsin has a real shit flag

16 Feb

I would have mistaken it for the work of a 1970′s print shop too.

!lilli!!lllliil!illilllillll!!i!lllll!!iillll…yes, syntax matters

14 Feb

Re Lisp without parens again, I dug up this old quote I clipped from a Reddit comment that sums up why Lisp’s parentheses are a real problem:

There is a reason all Lisp languages didn’t break through: the syntax is too monotonous. A program is written for humans to read, and humans are not too good with repetitive anything, including repetitive parentheses.

Another commenter writes:

People can go and love Lisp and [its] derivatives all they like, nobody I know finds deeply nested s-expressions very readable or writable. Syntax doesn’t matter much, but it does matter. I hacked on a fair amount of emacs customizations, but I always found it hard to follow the control flow.

People just don’t like parsing text like the title here for the same reason that they much prefer reading decimal or hex over binary. Sure, with time and experience, we can learn to deal with monotonous syntax, but given the choice, why should we put up with it? Some people get really good at sending and receiving Morse code, but Morse code is not only far less accessible than a keyboard, it’s unarguably less efficient: a proficient typist will always beat a proficient telegraph operator.

The parens of Lisp are repetitive noise which one can learn to cope with sufficiently well for some tastes, but not well enough for most programmers. Animvs solves this problem and goes a step further: by imposing a stricter indentation scheme and introducing first-class symbol highlighting, Animvs gives code a significant shape indicating structure that can be reliably parsed at a glance with low mental overhead.