Thinking is difficult

Ongoing twitter death throes and some issues in applied stats

Twitter is still slowly dying, and as part of its death throes, the real-time news function it used to have has been destroyed. A huge amount of what is retweeted across the site now is unverified speculation or misinformation. Fortunately at least we know it: people are documenting this phenomenon. If you have never been much on twitter, however, I want you to know that this has not always been true. It used to be different. Activists’ and journalists’ accounts used to be truly, properly verified; news sources all over the world used to genuinely put breaking news on twitter. But with the death of verification, that time is over. The twitter of @Alaa and the Arab Spring is gone. Today’s blue checks are trading on the currency of its memory. 

The problem is exacerbated by the fact that twitter’s algorithm systematically deprioritises any tweet which has an external link in it. So people who do want to put verified information on the platform (by linking people to the actual news sources) are punished for doing so. Such tweets go down like a lead balloon. No information in, no eyeballs out. An echo chamber forever. Elon, or whoever it was before him who made the decision to deprioritise tweets with links in them (and let’s not fool ourselves, this could well have been Jack), intends to keep you around on twitter using the traditional tactics of assholes and abusers: not by making it wonderful for you to go there, but by making it difficult for you to leave. 

(By the way, the best solution if you want to get true information out on twitter is to make a tweet with a link to a news source in it, and then quote-tweet yourself and amp up the quote-tweet. Truly, how complex are the behaviours that our dying ecosystem demands.)

Yeah, I know I keep insisting or predicting that Twitter is dying — and I’m right, it is dying — but it’s not fucking dying fast enough. And I guess I’ve appointed myself as the chronicler of its death foretold, partially because I’m still addicted to it. I have made I would say essentially no progress on that front. I’m not really planning to leave either, though I’m always tempted. The constant barrage of mostly-groundless speculation makes it pretty unpleasant to hang around.

It’s easy to get scared about a situation where misinformation is normal, and get very het up about people who spread misinformation, and probably it’s right to do so. But I think it’s also useful to remember that lying in public is not new; it has even, at times before now, been normal. I don’t think we should be too panicked or too harsh on ourselves about this either. I think lying is normal because knowing the truth is hard, and we do not do well as a collective with acknowledging uncertainty. 

And because knowing what is true is difficult, knowing what is right or good is difficult. In fact I think we can scratch even that and go one level up and also be more specific: even knowing what you yourself think is right is difficult. But I guess this is downstream of the fact that if you don’t try hard — and it’s very easy not to try hard since nobody else is going to reward you directly for doing it and in fact people would largely prefer if you didn’t — it’s actually hard to know what you think about anything at all. 

I think this has also has been made worse by social media. (I am pro social media and take no pleasure in reporting this.) There is now immense pressure to tweet responses to things as they are unfolding, which is to say, fast. In a first version of this post I tried to contend that such a pressure is fictional, but forget it — it’s real. Human behaviour is incredibly mimetic. To defy this takes extraordinary courage and attention. If you don’t pay attention you will mirror other people without you even realising it. People vary in the intensity of their mirroring, but it is rare to encounter someone who does not mirror at all.

So when it feels like everyone is sharing what they think on twitter — or at dinner, it also happens in analogue — in this kind of orgiastic rapid-fire emotional-discursive swirling vortex, we want to be in the rapid swirling vortex too with everyone, to be one of the brave and informed people, boldly sharing what they think. 

But are people sharing what they think? Or are they sharing some amalgamation of what their friends or family or people they admire think? It’s uncomfortable to even ask this, but we are not immune to propaganda, especially not from people we like and respect. Haven’t you ever heard yourself talking? Haven’t you heard other people’s words coming out of your mouth? I’ve felt it, and I hate it. This danger is present whenever you are in a situation where people are invested in you thinking what they want you to think. It’s not just that these situations make you afraid to speak your opinions, though there is that. But it’s worse than that, I think. These situations make it difficult to even know what your opinions are. 

Well, so what, you might say (to yourself) — how on earth could this be difficult — don’t you know what you think? 

But how could you know what you think if you don’t sit down, in the quiet, possibly in the dark, almost certainly alone, and think it? 

Compulsively unspooling every random thought that comes into your brain out of your mouth does not constitute sharing your opinion.¹ The stream of thoughts that comes unfiltered into consciousness is not always useful or indicative of anything particular inside you; it is not even always what “you” “think”; it can be snippets of tv news or snatches of reels or twitter, stuff you heard and saw elsewhere, things other people said that are going to come into your brain and then out of your fingers or your mouth. It’s practically automatic. But that is not how real thinking gets done.

***

One thing that I think would surprise outsiders to academia is just how much even academics want to avoid thinking.

I always remember one of my professors in grad school² saying in class that econometricians have more in common with fashion editors than either party would be willing to admit. This was shocking to the class then and it’s shocking to people when I repeat it now. Like, forget ragging on applied economists or policymakers for being faddish and suggestible — econometricians have this problem. Research, even in abstract mathematics, has trends. Fields and groups like to have things done the way they do them. That doesn’t mean that research is bad or useless, or that it is unable to find truth, but it will find truth in spite of these tendencies and because other forces countervail them. And the more we are aware of this trend-driven aspect of our minds, the better at econometrics we will be. 

I actually think one of the great gifts of going to grad school (this is not a recommendation nor an endorsement) is that it gives you this great amount of time for thinking. I mean you pay for it in foregone earnings and probabilistic exposure to exploitative conditions, but that time is really valuable. For a certain type of person it’s a good tradeoff. I believe unstructured time is an essential input into thinking. As life goes on it becomes more and more important to protect that time.

Funny things happen if we don’t protect it. One of those things is that we experience ourselves as more stupid. Not having time to devote to understanding things makes them seem far more difficult than they are, because it contributes to this feeling that understanding must come quickly. And then I might feel that if I am not quick in understanding, then this must mean I am stupid, or that something is going wrong. 

But in fact the opposite is nearly always true. I learned over many (many, many) years of school that if I think I understood something well in the lecture there’s a solid chance I simply confidently misunderstood it. If you struggle with a concept, at least you are avoiding that trap — you are actively engaging with it, and the struggle is usually fruitful and if not then interesting. Overall, that’s a good process. Learning something new feels for the most part like confusion; confusion is the way that thinking feels. Clarity is the end product. It is not the process itself.³

I also think that a lot of what we mean when we say math or physics is difficult is that it produces this feeling of confusion for a long period of time, and requires lots of dedicated, focused attention to learn and thus resolve it. Of course, being really good at textual analysis or French also takes time and is also very difficult, so I don’t know why math specifically gets this bad rap except perhaps that it’s held to be more universal than French and that the symbols are intimidating. But the letter “a” was also intimidating to us once. Since then we have grown in pride and do not like to feel intimidated. That is what hampers our learning. 

I have thought about this a lot over the years regarding the (somewhat) general perception that Bayesian statistics, my own area of expertise, is somehow harder to learn and use than frequentist statistics or machine learning or other alternatives. While anyone would of course feel a little pride in being accused of having mastered something complex, in fact Bayesians usually reject this charge because it is levelled by people opposing the widespread use of the method — it is seen by these critics as “too hard” for general use among researchers and scientists of all kinds. Setting aside the fact that many applied researchers have learned to use Bayesian methods, I, like most Bayesians, contend that the alternative approaches only appear to be easier to use; that they are overly simplified as they are currently presented, and that by hiding their own areas of genuine complication, these alternatives are more open to being poorly or mendaciously used.

When asked to defend the method’s difficulties, we Bayesians often conceptualise this situation as: “Bayesian inference is hard in the sense that thinking is hard”. I recently looked this phrase up and it is a slight misquotation, but the real quote actually has a specific attribution: it comes from the biostatistician Don Berry. In this excellent article on why there are so few elementary Bayesian statistics courses, he cites the general feeling that Bayes is “too hard” for most undergraduate students as one factor, and elucidates this as follows:

“Among the students who have the most trouble are those who cannot unlearn an attitude developed in high school mathematics courses: namely that solving problems means plugging into formulas — thinking is strictly optional. Bayesian statistics is difficult to the extent that thinking is difficult. Bayesian statistics is demanding to the extent that the instructor requires that students think.”

You kind of can’t force students to think, though. You can only set it up so that thinking is in their interest. Even that is actually not that easy to do! Setting an assignment or an exam that truly requires students to think in order to solve the problems is really not that easy. Every year I try and this also forces me to think hard. And, I must stress, I am disincentivised from doing this. Nobody wants me to set truly challenging and interesting exams. 

You are by now used to listening to ambient complaints about how students don’t want to think anymore or whatever. I don’t like to see the “anymore” tacked on there. It’s true that most students do not want to think but that is because students are people. Most of the faculty and administration does not want to think either. Students have every right to bristle at this complaint when it could not be clearer that we, too, do not want to engage in thinking and for the most part we are happy to be in environments which are inimical to it. We have the right to observe this tendency in our students, but we do not have the right to criticise it. 

Of course if I admit that I am no better and no more willing to do serious thinking than my students or anyone else then it makes me feel a lot less good about myself. Or more specifically, it makes me feel a lot less certain of what I believe and what I know. Maybe this tendency to shy away from thinking is just regular old discomfort with uncertainty. And perhaps the hallmark of real thinking, the thing that distinguishes it from regurgitation or rationalisation, is our inability to predict the outcome of the thought.

This has to be part of why thinking is so unpleasant and therefore difficult. If I acknowledge the possibility that I do not yet know what the answer is, and that perhaps I do not even know what I feel or think about some moral or political topic, then I am by definition leaving myself open to painful revisions in my worldview. Like, if I don’t know what stands between me and understanding calculus, but it is resisting my attempts to understand it, it is likely that much of my current intuition-based understanding about this part of the world (er, so, functions) is either scrambling it or blocking it — so in some sense, at least in some certain cases, my previous intuition is wrong.

For most of us, even being open to letting go of our current understanding is uncomfortable. Admitting to the possibility that we are wrong or even just ignorant is difficult, and that defines learning. But the admission itself makes us feel in some way exposed. I wanted to end on some kind of positive note, but unfortunately I think there isn’t one. Thinking is just really difficult. 

1. Bad news for me personally!

2. Just as a reminder it’s my policy not to name any names in these personal stories.

3. I’m not going to do this as a general policy but in this case I am going to cite one of my own tweets which says this: Something curious happens in research which is that you generally also have to move through the point of clarity, until a new level of confusion is reached, at which point, you can know you have really understood it even further than the initial clarifying point. There is actually an incoherence in the center of everything. Or something like that. (Here’s the tweet.)