Traffic Lights and Modern Epistemology
If you prefer to listen rather than read, this blog is available as a podcast here. Or if you want to listen to just this post:
The other day I was on Hacker News, and someone had taken advantage of the Ask HN feature to inquire where they could go for dispassionate discussion now that Slate Star Codex was gone. I have tried to go back and find that post, and I couldn’t, so I may be misrepresenting some of the details here, but if my memory is correct the top comment was by someone who made the point that a dispassionate discussion of something could only occur if it didn’t involve the people affected, which was exactly the wrong way to have a discussion and what has been happening with too many social issues up until this point. Implied further in this comment was the idea that dispassionate discussion was the wrong way to solve a problem.
This comment brought up an important and necessary point… Up until the moment where it started talking about solutions. Certainly you wouldn’t want to exclude the people affected by an issue from a discussion of that issue. Even if, and perhaps especially if, the issue made them angry, and the discussion ended up not being dispassionate because of that anger. But at a certain point, what we really want to do is solve the problem in the best fashion possible, which requires objectivity, and yes, some dispassionate discussion. In other words it may be perfectly justifiable for people to be angry; it may be and probably is important for them to have their say, to explain exactly why they’re so angry; but in the end anger is rarely the best strategy for solving the problem. In fact, if you allow the most angry to dominate the discussion, you're far more likely to end up with a really bad solution than the best solution.
Perhaps an analogy would help to illustrate what I mean. To leave my neighborhood and head south I nearly always have to stop at a particular traffic light. On occasion I end up waiting at this light for what seems like forever, because it's heavily biased in the other direction. As the minutes drag on (full disclosure: I believe the longest I’ve ever waited is a hair under three minutes) I get understandably annoyed, and sometimes, if I’m already in a bad mood, by the time the light changes I’m pretty angry.
The other morning that's exactly what was happening, I was waiting at this light for what seemed like a very long time and getting increasingly annoyed at it. But this time I noticed something, after all that time when it eventually turned green, there were only two cars waiting, me and another guy opposite me, while during the time I had been waiting many cars had passed in the other direction. Which led me to wonder if perhaps, when considering all the traffic that passed through that intersection, if the system made sense. This was actually not the first time I had had this thought (though it was the first time I noticed how meager the traffic was on my street) but it’s easy to forget the system as a whole when you’re being inconvenienced by one part of the system.
But what does it mean for the system to “make sense”? Or to consider my specific case, I was angry at this light because it was constantly causing me discomfort, but I had never really engaged with the question: what system should be used to calibrate that light?
Some possibilities:
Would you calibrate it based on time of day? (Indeed if you show up before 6:30 the light just automatically changes as soon as someone approaches the intersection, and I definitely prefer the system in operation before 6:30 to the one after.)
Would you base it on what the city council felt was fair? Perhaps take a vote on the calibration of every intersection? Maybe even expand that vote to everyone?
Perhaps, rather than try to optimize every intersection you might just place every street into one of three buckets based on the level of traffic, perhaps high, medium and low, and then categorize intersections based on a matrix. The intersection of two high traffic streets would get one setting, while the intersection of a low traffic street with a high traffic street (my intersection, presumably) would get a different setting.
Would you measure traffic in each direction? Track the time each car had to wait, add it together and try to make the two directions equal? (i.e. If I have to wait for three minutes then that's fair if it equals 6 cars waiting 30 seconds in the other direction).
Similar to the foregoing but with a different metric, you might be trying to minimize the pollution generated by idling cars, and while time spent waiting would be part of it there might be other factors as well, like managing how many cars are accelerating after being stopped.
Finally, we might use anger as our guide for adjusting the light. Perhaps the method just mentioned of aggregating the wait times in both directions never makes the people going the other direction annoyed, but makes the people going my direction furious. You might try to optimize for lowest aggregate anger, and find that you could make people going the other direction stop more often without an appreciable increase in their anger but with a significant reduction in my anger.
Beyond the methods mentioned above there are still other standards I didn’t mention, for example I had a co-worker many years ago who was convinced that businesses paid the city to increase stoppage at nearby intersections as a form of advertising. The point being that, while it would certainly be unfair if I had no say in how this traffic light worked, the problem of calibrating even a single traffic light is pretty complicated one. It may be that when you really dig into things, the best system might end up being one which causes me quite a bit of delay. And as you can see, even deciding on the system to decide if a delay is justified is complicated.
All of the foregoing falls under the heading of epistemology: the investigation of what distinguishes justified belief from opinion. Am I justified in believing the traffic light is unfair, or is that just an opinion, how do we define what’s fair? I might prefer it if the light just changed as soon as I approached it, but that system is almost certainly not optimal during rush hour. I might accept that total aggregate time is a good way of determining fairness without necessarily being happy about it, and depending on the evidence, I’m sure I could be talked into a road classification system of three categories, on the basis that the measurements required for other systems are difficult to make. In fact, I’d probably be okay with any of the methods I mentioned (not the businesses paying for stoppage one of course, and the aggregate anger one probably leads somewhere bad as well). Of course, part of being this magnanimous is that this issue is very low stakes.
But what if the stakes were much greater? What if I was convinced that one method for determining the length of a traffic light increased the chances of me dying at the intersection? Or that another method might cut economic growth in half? I might be far more invested in how this decision was made, and far less likely to accept any old system. And it gets even worse when all methods have seriously bad outcomes and we’re being asked merely to choose which bad outcome we prefer. I might choose a small increase in the fatality rate over halving economic growth, whereas someone else might make the opposite choice, and assume I’m a horrible person for putting people’s lives at risk.
Of course this is not a new problem, rather it’s a very old problem, and in the past intractable problems have been solved by things like war, enslavement, dictatorial powers, and just about any other injustice you can imagine. But over the last several centuries we developed some tools for avoiding the worst of those injustices. Things like freedom of religion and speech, widespread democracy, rule of law, etc. Now I’m not suggesting that any of these things are free of flaws, they are in fact riddled with them, but before we cast them aside it’s important to remind ourselves how bad it was historically.
At this point it would probably be beneficial to talk about epistemology using examples of things people are actually getting worked up about, rather than the example of a nearby traffic light. In doing this I hope to pick topics where the differences in opinion are easy to see, but not so great that they overwhelm the discussion, I’m not sure I’ll succeed in this, so I ask for some patience as I proceed. With that said let’s look at some current events through the lens of a few different epistemological frameworks.
One well known framework that seems to be getting a lot of attention these days is the one Scott Alexander described in his post Conflict vs. Mistake (link goes to an archive.org version since SSC is still deleted, and the NYT still hasn’t published its article). It’s a pretty good post and you should probably just read it, but for those that don’t. It describes two ways of viewing political struggles, mistake theory and conflict theory:
Mistake theorists treat politics as science, engineering, or medicine. The State is diseased. We’re all doctors, standing around arguing over the best diagnosis and cure. Some of us have good ideas, others have bad ideas that wouldn’t help, or that would cause too many side effects.
Conflict theorists treat politics as war. Different blocs with different interests are forever fighting to determine whether the State exists to enrich the Elites or to help the People.
Clearly my description of the traffic light issue is a description from the perspective of mistake theory. (Though the aggregate anger methodology comes close to conflict theory.) That there is some optimal way to time traffic lights, and we just need to figure out what it is, that there is no war between people traveling north and south at the intersection and those traveling east and west. But of course when you scale things up, things become a lot more muddy, which takes us to the subject of statues.
As I believe I mentioned, one of the statues which got torn down was that of an abolitionist who fought and died in the Civil War. People operating from mistake theory will describe that as a mistake and go on to identify similar mistakes on both sides of the issue. It was a mistake to put up statues to Confederate generals. It was a mistake (a very bad one) when George Floyd was killed, so if we can identify what statues are mistakes and which are not, we’ll take the former down and leave the latter alone, and if we can identify the policies and training and culture which lead to Floyd’s death we’ll fix those too and eventually we’ll be able to put the whole issue to bed.
On the other hand, from the conflict theory side of things focusing on mistakes is just a way of getting back to the same crappy status quo as soon as possible. Of deflecting the discussion away from systemic racism into a discussion of whether people went too far when they tore down the statue of Frederick Douglass. Of a path that leads to a few tiny reforms, but that basically keeps the same corrupt police around doing the same awful things. Conflict theory would go on to say that TV networks didn’t do dumb things like remove episodes of Golden Girls, Community and 30 Rock, because they’re combatting racism, they did it because they’re obviously on the side of the elites, and doing that deflects attention away from real grievances to trivial ones. And finally, that it doesn’t matter which statues get torn down, because tearing down statues is a great way of showing passion, and passion is the only thing that’s going to sustain the unity of the oppressed long enough for them to get what has long been denied them.
When considering this dichotomy of mistakes vs. conflicts, it’s hard for me not to see the world through the lens of mistake theory, and I think most of my posts, including this one, naturally proceed from that epistemology. But in my more pessimistic moments it seems obvious that at some fundamental level it’s all about conflict, and always has been, and that the enlightenment tools I mentioned earlier, like freedom of speech, etc. were just exceptionally clever ways of masking the conflict, or that they contained the conflict, but only temporarily. Or perhaps they represent a Noble Lie, an ideology that is fundamentally untrue, but which works to maintain social harmony.
As something of an aside, it’s interesting to note that you can see this epistemological split in the political parties, and it appears to be widening. On the right clearly the Trump/alt-right branch are the conflict theorists, and the Mitt Romney/Neo-con branch are the mistake theorists. While on the left Obama/Biden/Clinton are largely mistake theorists, while Sanders and the people currently protesting are conflict theorists. I couldn’t say what this means for the country as a whole, but it’s probably bad.
Mistake vs. conflict is not the only way of looking at things, though it covers a lot of territory, and the next framework I describe may just be a subcategory of it.
There was a time, and I’m old enough to remember it, when the history of the country was pretty sanitized. People who talked about Washington didn’t mention his slaves, and when discussing JFK you didn’t mention his mistresses. Manifest destiny was the obvious next step in the progress of the nation, and the cowboys were always the good guys. I want to call this an epistemology of national greatness. That what was true, or at least what was emphasized were those things that made the country and its history look noble, and of course this took in all the things that led to the formation of the US, so Columbus is obviously a great figure with statues and holidays, not the first in a long line of bad Europeans.
Of course this way of determining truth or what to celebrate and emphasis isn’t particularly scientific, or empirical. And so at some point in the last century (almost certainly before I was in school, but these things take a while to trickle down) that narrative switched to an emphasis on not only uncovering the bad things, but emphasizing them to make up for lost time. Of making sure that when you talk about Jefferson his affair with Sally Hemmings has a prominent place. That when talking about westward expansion you discuss the Native Americans and broken treaties for at least as much time as you discuss the settlers, and so forth.
What’s interesting about this, and really why I’m talking about this framework at all, is that it turns out it’s much more difficult to achieve agreement and unity under this system than using the national greatness system. You would think that by really emphasizing facts, that it would be easier to get people on the same page, but it’s actually harder. Under national greatness, if someone or something was important in the history of the country then they’re worth celebrating. It’s a narrative that’s self reinforcing. We celebrate important events and people. How do we know they’re important? Because we celebrate them! But once you pivot to facts you can generate all manner of narratives.
How does the fact of Jefferson’s relationship with Sally Hemmings relate to the fact that Jefferson was the author of the Declaration of Independence? Oh, and are you 100% sure Jefferson did have a relationship with Hemmings? Under the standard of national greatness the narrative is easy, Jefferson was important in the history of the country so we celebrate him, and put up statues and monuments to him regardless of his failings, which we either ignore or mention in some footnote. Under a standard of using the facts to determine whether we should celebrate Jefferson, we can come up with at least a dozen narratives, and each one has a different recommendation for what to do with the Jefferson Memorial. And to be clear I’m not saying this is bad, I prefer to get the facts out, but when you compare the America of today with the America of, say, the 60s one of the big differences is the shift from a patriotic, national greatness epistemology to this one.
As one final thought before we move on, I’ve never quite understood why the North was so willing to spend massive amounts of blood and treasure to prevent the South from seceding. But just now I reconsidered it through the epistemological framework of national greatness and it clarified things in a way that nothing previously had, which is not to say I don’t still have questions, but viewing it through that lens was very illuminating.
The final framework I want to consider is the one I mentioned in my previous post, If We Were Amusing Ourselves to Death in the 80s, What Are We Doing Now? And rather than rehashing it in its entirety, I’m more interested in taking another crack at answering the question what are we doing now? For those that haven’t read that post it was a discussion of the book Amusing Ourselves to Death: Public Discourse in the Age of Show Business by Neil Postman, published in 1985. Postman’s thesis is an extension of Marshall Mcluhan’s observation that the “medium is the message” and boils down to the idea that there is a “connection between forms of human communication and the quality of a culture.” For Postman, culture was of a higher quality when communication largely took place via print (newspapers, books, etc.) and that it took a dive in quality with the introduction of the TV. From the book:
[U]nder the governance of the printing press, discourse in America was different from what it is now—generally coherent, serious and rational; ...under the governance of television, it has become shriveled and absurd…like the printing press, television is nothing less than a philosophy of rhetoric. To talk seriously about television, one must therefore talk of epistemology. All other commentary is in itself trivial.
One can grant that the dominant medium of communication strongly impacts epistemology without necessarily granting Postman’s further claim that it’s all been downhill. Indeed I find it hard to imagine how anyone could deny the effect of the medium of social media on our current epistemology. The question I continue to grapple with, is what are those effects? I’m starting to feel pretty comfortable declaring that they’re, on net, bad but the specifics of their “badness” is something I’m still working through. I have high hopes for the grandstanding theory, which I encountered the other day on a podcast, but I’ve yet to read the associated book. I’ll report back when I do. However, it does seem certain that if nothing else, social media has fractured epistemology and discourse. That under national greatness there was obviously only one thing to do with the Jefferson Memorial, that as people started focusing more on Jefferson’s failings you can imagine the options splitting into three, to be decided by congress, leave it alone, add some additional plaques to explain things, or tear it down, and to be honest the third one would never get serious consideration. But currently I’m sure there are at least a dozen proposals, ranging from ringing it with an alt-right militia, to replacing it with a statue of Sally Hemmings, and everything in between.
There’s a quote I keep coming back to in this blog, from Chief Justice Oliver Wendell Holmes:
Between two groups that want to make inconsistent kinds of world[s] I see no remedy but force.
I’d like to think that this statement is actually not true, that over the last few centuries we have developed other remedies besides force. Freedom of conscience (of which freedom of speech is a part) was, I believe, particularly successful. (We’re not immediately going to go to war over differing beliefs.) Solving things by voting on them was also a major step forward. But it’s interesting how, beyond all of the other ways in which these tools are under attack, they just don’t work nearly as well when you end up with more than two or three sides. When the narrative has fractured into dozens of pieces, as appears to be the case at the moment, these tools become more difficult to use. Taking free speech as an example, even if sensible suggestions are being made somewhere by someone how do you find them amongst all the yelling? And this is without the additional problem of free speech increasingly being seen as outmoded and a tool the majority uses to silence the oppressed.
As far as voting, that works pretty well if there are only two sides. Making deals involves only two parties, and even if you can’t make a deal your side will eventually be in power and waiting your turn seems preferable to bloodshed, that’s not the case when you’re a member of one of 30 factions, also how much voting do we actually do on the most contentious issues of the day? Certainly the vast majority of social issues have never been voted on. Voting can be a tool for remedying inconsistent worlds, but you have to use it first. And added to all of this, everything increasingly seems like a zero sum game.
I feel like it’s safe to say that no one is clear on where things are headed, or that it will inevitably be bad because discourse has moved to social media, but when you tie all of it together, toss in a profusion of conspiracy theories, and an exceptionally divided country, I think what can be said for sure is that from an epistemological perspective, we’re in a very weird place.
It’d be nice if things were as simple and straightforward as agreeing on a system for the traffic light. They’re obviously not, but nor do I think they necessarily need to be as complicated as we’ve made them. I think it’s easy to forget how much progress actually has been made over the last several centuries, and it’s even easier to forget how fragile that progress is.
Is it true that you should donate to this blog? What standard would you use to decide whether it’s a justified belief as opposed to just my opinion? Well, there is another framework I didn’t mention, that you uncover the truth of something by doing it. Maybe you should give it a try.