Sunday 8 November 2020

You are gaslighting yourself.

I am a mathematician. Not professionally, but mentally. I'm one of those people who had an affinity with numbers and logic from a very early age, which led eventually to a degree in maths and a career in IT. I understand that a lot of what seems bleeding obvious to people like me is opaque to many people. (This is not to be boastful: we all have our strengths and weaknesses, and Lord knows there's plenty I don't understand, like football and small talk and not making people want to punch me.) But I honestly did not appreciate until this week the extent to which obvious mathematical truths are complete mysteries to such a huge chunk of humanity.

Of course I'm talking about the cursed election. You may have seen these graphs:

Those graphs are evidence of something highly suspicious happening. Clearly and obviously, to anyone who understands numbers. That has nothing to do with what they're measuring. If they were graphs comparing the performance of two brands of dishwasher, they'd be suspicious. If they were exchange rates or share prices, the financial regulator would be demanding an explanation from the banks involved, and actively considering raiding those banks if such explanation was not forthcoming. The idea that those numbers are not suspicious is just preposterous. I shan't bother explicating why; one thing I've discovered this week is that those who can't already see it will determinedly continue not to. But the evidence honestly couldn't be much clearer. And yet the dominant claim, repeated all week throughout the world's media, is that there is "no evidence". Not that the evidence is, on balance, unconvincing, or that it can be explained, but that there simply is none. That so many millions of people are apparently willing to believe such a thing is the most damning indictment I have ever seen of the state of maths teaching.

Now, let's be clear here about what the word "evidence" actually means. It doesn't mean "proof". When a defendant is found not guilty, all the evidence against them doesn't cease to exist; it is still evidence against them, which has been weighed and found not damning. Even though they were not guilty, the evidence against them is the reason why it would have been remiss not to try them in court at all. Evidence is something that is worth looking into.

There are various scenarios that could explain graphs like that. "This is normal; nothing to see here" is not one of them — but that's the one we've been given, again and again, smugly and condescendingly, by people insisting that even being suspicious of numbers like that is a sign of knuckle-dragging stupidity.

But, utterly unsurprisingly, it turns out that the evidence was worth looking into. For instance, Antrim County, Michigan discovered that software had allocated 6000 votes to the wrong candidate. That's not a conspiracy theory; it's not a paranoid accusation by the losing party; it's a fact, verified and confirmed by the election officials. The numbers were evidence that something was wrong, so people investigated the numbers, and found that indeed something was wrong. That's how evidence is supposed to work. It may be that the software was hacked with the intent of fraud; it may also be a mere innocent error. But what it most certainly is not is nothing.

That software, incidentally, is in use in 47 counties. Since we now have proof that it malfunctioned to the tune of thousands of votes in one of those counties, that proof is in turn evidence that it may have done something similar in the other counties. That evidence should of course be investigated. It might turn out that it only miscounted in Antrim and worked correctly everywhere else. That's how evidence is supposed to work too. But claiming that it might have behaved the same way everywhere it was used (which is, after all, what such software is explicitly designed to do) is not crazy, is not sore losing, and is not an attempt at a fascist power-grab. It's an entirely reasonable suggestion. And refusing to investigate it would be insane.

I currently work in financial regulatory reporting. If I were to see numbers like that and not investigate them, I could go to prison. And I'd deserve it.

The graphs are far from the only evidence. Larry Correia set a load of it out here, along with a good explanation of what a red flag is in an audit and what it does and doesn't mean. I recommend reading it if you're interested in the US election.

But, believe it or not, I don't want to talk about the US election.

I've deliberately not mentioned Trump till now, because this isn't really about him — or shouldn't be. Trump will be gone in four years or a few weeks, and either one is the blink of an eye. (Note to teenagers: Yes it really is. Just wait.) Democracy, we should hope, will be with us somewhat longer.

Four years ago, in a context that was different yet ultimately the same, I wrote this:

But to think that that makes the theft of our rights and powers OK is to fall into the usual trap of thinking that democracy is just a decision-making mechanism, and that therefore it is the decision it reaches that matters. But democracy is not primarily a decision-making mechanism. I mean, really, if you were setting out to design a good way of making good decisions, would you come up with democracy? Of course not. Because it's laughably useless.

However, democracy is a very very good civil-war-prevention mechanism.

Democracy works because enough people believe that they will get their way some of the time. Because of that, they are willing to accept the result when they don't get their way — they know their time will come soon enough. It is this general belief spread throughout the populace that has effectively stopped people resorting to violence. This is why not only actual propriety but the appearance of propriety in the electoral process matters so much. This is why it is vital that suspicious events be investigated seriously. This is why investigating events and finding them to be above board is not the same as pre-emptively declaring that they are not worth investigating in the first place. The process matters far more than the result, because the process is the alternative to what our ancestors did. Our ancestors killed each other. In huge numbers. The best estimate of the effect of the English Civil War on Ireland is that about 41% of the population were killed. No, that's not a typo. Over disagreement about who should be in charge of another country.

Personally, I prefer voting.

American ideas have a way of spreading. For a long time, that hasn't happened with their electoral fraud. They have generally had a bigger corruption problem than the rest of the English-speaking world for some reason — just compare the number of entries for the US to those for the UK, Canada, Australia, and New Zealand on Wikipedia's list of controversial elections. That is about to change.

Because any tactic that works will be used.

Try and forget about your dislike of Trump for a second. Sure, you want him gone, for whatever reason. It ought to be possible to say so and yet still notice crap like this:

Just a general observation from a foreigner feeling ever more foreign since Wednesday morning: I have spent election night in many countries over the years, and have never seen what I saw on Tuesday night. And I am amazed that even the parochial brain-dead American legacy media could pass off what happened in Philadelphia as normal. Everywhere else the polls close and the riding or constituency counts the votes until they're all done and they have a final 100 per cent result - by 9pm, 11.30pm, 3am, however long it takes. But, in Philadelphia and elsewhere, they suddenly stop counting and everyone goes home until late the following morning to start counting the boxes that have shown up under cover of darkness.

And everybody on ABC, CBS, NBC pretends this is perfectly normal.

If that were to happen in a British election, we'd regard it as completely mental. I'd expect every candidate to protest — not on the grounds that they suspected their opponents of nefarious shenanigans, but for the more basic reason of seriously, what the utter fuck? This is just not how one does elections.

Until this week.

Because what happens when you insist that this is unremarkable, that there's nothing to see here, that this is a correct way of conducting affairs, is that you make it more likely. Half of America have their own reasons for insisting that it's AOK. You don't have to agree with them to prove your hatred of Trump. As long as the British were outsiders looking in, regarding such behaviour as a bizarre foreign American thing, it remained a bizarre foreign American thing. This week, when you declared you were fine with it, you invited it in.

Political parties are full of devious cheating bastards — because they are groups of human beings, so of course they are. Devious cheating bastards will do whatever they can get away with to win — you could look it up. Up till last week, there were things they thought they couldn't get away with in Britain. You are now busily telling them otherwise. You have told them that you're happy for polling centres to have their staff sent home and for tens of thousands of votes to be delivered while they're away. You have informed them that you will avert your eyes from obviously suspicious numbers. You have even told them that you will accuse anyone who does notice obviously suspicious numbers of corruption. The safeguard against this stuff was your refusal to stand for it. And you threw it away.

Any tactic that works will be used. Sooner or later, it will be used against you. Once these practices take hold, they'll be in our elections in twenty years, in fifty years; they'll be entrenched in the elections your grandchildren vote in. They won't even have heard of Trump, but they'll be stuck with this shit. Until they get fed up with it not working, so start killing instead.

Because half our generation eagerly lied to themselves because they disliked one politician, who wasn't even in this country.

Trump didn't do this. You did.

Update, 10/11/20:

The original version of this article contained a third graph. It has since been removed from its original source, so I've removed it from here too.

Monday 6 April 2020

On the occasion of Boris Johnson entering intensive care.

If you are so dedicated to the grand cause of tribalist politics that you are welcoming the prospect of a child being born to its father's widow, then your soul is fundamentally broken, and your presence is a poison to all who encounter you.

That is all.

Wednesday 18 March 2020

Covid-19, lightning strikes, and the nature of risk.

There's this way of measuring risk that's so popular it's a cliche: "You're more likely to be struck by lightning than [insert supposedly unlikely event here]." It's interesting that we're hearing it used less these days about terrorist attacks — in the UK, you have actually been more likely to be hurt by a terrorist than by lightning for some years now, and perhaps, even without looking up the figures, people are intuitively sensing that. That rather annoying thought aside, what I really want to discuss here is the fact that the comparison is a big steaming pile of fallacy.

Because the way the expression is used is always to express derision at the very idea that anyone might worry about such a tiny risk. The logic goes:

The probability of X is lower than the probability of being hit by lightning;
the probability of being hit by lightning is extremely low
: therefore taking measures to avoid X is a stupid waste of time.

But here's the thing. In the Developed World, every tall building has a lightning conductor on top, and so do most of the smaller ones. And every crane. Every car has been designed to protect its occupants in the event of a lightning strike. Aeroplanes are designed to survive lightning strikes. The electrical ring main in your house is lightning-proof. The entire national electricity grid has lightning resilience built in. Scientists and engineers keep researching lightning, so that we can keep improving our lightning-proofing measures. And we teach our children not to shelter under trees in thunderstorms. Apart from that last one, this all costs money — money and time and resources and effort. Our entire society is, at great expense, largely lightning-proof. The reason we've done this is obviously not because it's such a rare event that it's not worth doing anything about.

So the logic is actually:

The probability of X is lower than the probability of being hit by lightning;
the probability of being hit by lightning is high enough that we collectively spend billions on mitigation measures against it
: therefore taking measures to avoid X is a stupid waste of time.

It's even rarer than a thing that isn't rare!

I am reminded of this by the scaremongermongering fuckwits out there still, even now, saying that Covid-19 is not much worse than the flu.

Influenza is a big deal. We research the new strains constantly, so that we can identify which are the ones we need to vaccinate against each season. We develop a new vaccine for those strains every year. We deploy that vaccine to millions of people. We prepare a certain amount of capacity in our hospitals to deal with serious flu cases. We train doctors and nurses in its treatment. We obviously do not do all this because flu is not worth doing anything about.

Again, here's that logic, only even worse this time:

Covid-19 is not much worse than the flu;
the flu is a serious enough problem that we pour massive resources into combatting it
: therefore making all this fuss about Covid-19 is stupid.

It's only slightly more dangerous than a very dangerous thing! Genius.

Part of the problem here is the word "fuss" and its various synonyms — "worry", "panic", etc. These are loaded words, all being used to mean "effort", but with derision of that effort built in. It's true: we don't make a lot of fuss about flu; we don't worry or panic about flu; but we do put lots of effort into fighting it. We just don't think of that effort as fuss or worry or panic, because it's routine. But it's still effort, and using loaded words doesn't create some logical difference between that effort and the effort put into fighting Covid-19.

And of course the other glaringly obvious point is that we aren't replacing flu with Covid-19. We're getting both. This is a new risk piled on top of all the old existing risks we already know about. Very few people do think about the risk of dying of flu — or tuberculosis or hantavirus or an embolism or a car crash or a freak shopping-trolley accident (it happens). What we do think about and take into account in the back of our minds is the risk of dying in general. That risk just suddenly increased, so of course we're seeing an adjustment. And it's not surprising or irrational that that adjustment is so extreme when the size of the risk is currently unknown and unknowable.

Now, there is an argument to be made that, whilst Covid-19 just increased the risk of dying, we have, on the other hand, been steadily decreasing the risk of dying for the last few decades, and so we might rationally offset those two factors: sure, things just got significantly more dangerous, but cancer is no longer a death sentence, you can live a long life with HIV now, and modern seatbelts and airbags are frankly amazing, so maybe this stuff balances out. Thus far, I have heard that reasonable argument from precisely zero people.

And then there's the fatality rate. The scaremongermongers measure Covid-19's badness by its fatality rate, with the underlying assumption, sometimes implicit, sometimes explicit, that those who don't die will be just fine. Sorry, but no. There's a whole world of mostly horrible grey areas between dying and being AOK. Covid-19 causes pneumonia. Pneumonia is nasty. It's very painful. It damages your lungs, usually long-term, often for decades, often permanently. "Recovering" from pneumonia could well mean that every respiratory illness you get for the rest of your life will hit you harder than it would have otherwise — and there are a lot of respiratory illnesses out there. Imagine if, every time you caught a common cold, it turned into a chest infection. Imagine if you got so many chest infections, you built up a resistance to most antibiotics. That's what a significant proportion of survivors are facing. And then some lucky sods can carry Covid-19 completely asymptomatically. It's not all that bad for everyone, but it's a lot more complicated and worse than just putting survival in the "no problem" column.

Another thing you've no doubt been hearing is that Covid-19 mainly kills elderly people. This is true, and, not to be callous, but, frankly, after a certain age, whatever you die of, you really die of old age. When she was merely in late middle-age, my grandmother broke her leg and didn't notice, continuing her daily ten-mile walks across the Yorkshire Moors while being vaguely annoyed that the limp wasn't wearing off. In her nineties, she broke her leg and, though it took a year, died. We know this is how life goes, and are generally resigned to it. But perhaps you haven't fully thought through what "elderly" means in the context of Covid-19 fatalities.

You don't suddenly drive off a statistical cliff-edge at the age of seventy. There's a curve. Yes, at seventy, you're at much higher risk. At sixty, not as bad, but still higher risk than a fifty-year-old. And so on. I'm forty-five. I certainly don't think of myself as elderly. But I'm on the curve. I'm unlikely to get away from Covid-19 as cleanly as a twenty-year-old would. I run part of that increased risk that shows up in the stats and gets described with the shorthand term "elderly". Maybe you do too.

I'm not panicking about this. I half hope I catch the damn thing, so I can grow some antibodies and cease to be a danger to my wife. But only half, as it will probably be fucking agony. But this is not panic, it is not fuss, it is not a ridiculous overreaction to nothing much: it is a reasonable response to a rational analysis of something that really is much worse than flu.