Follow the reluctant adventures in the life of a Welsh astrophysicist sent around the world for some reason, wherein I photograph potatoes and destroy galaxies in the name of science. And don't forget about my website, www.rhysy.net



Tuesday, 28 April 2020

Thing What I Was Wrong About

Any blogger with any sense will tell you that there's a risk of saying stuff that's later found to be utter crap. It may seem perfectly sensible to go on a good angry rant about how Jar Jar Binks was the worst character ever, but then you learn that the actor was traumatised for life by the deluge of hate mail and suddenly things don't seem quite so much fun any more.


And yet in science we're supposed to delight in proving ourselves wrong. Indeed, everyone wants to overturn Einstein, or discover a new and unpredicted particle. No-one wants the stinky old Standard Model to be entirely right, although pretty much everyone wants their own special patch to be basically correct.

This peculiarity is so at odds with everyday life that it may explain a lot about the mess the world is in. Science progresses by making mistakes, whereas in politics an open admission of a mistake is tantamount to publically slitting one's wrists while falling into a tank of flaming piranhas who are all infected with COVID-19. Figuring out the general conditions as to when it's fun to be wrong (and when it isn't), and how we can apply these lessons in the wider world, will be the subject of a future post.

As a sort-of prelude to that, what about lil' old me ? A few years ago I chronicled how some of my major beliefs have shifted. But blogging makes it possible to track one's mistakes as never before ! So instead of evaluating past mistakes, this post will be an active attempt to see if any of my current beliefs are wrong. What, then, have I prophesied that turned out to be correct, and what utter garbage still haunts forgotten corners of the web ? You might have seen me rant a little on Decoherency that I think such an exercise is badly needed for popular political commentators, so this post is my effort to put my money where my mouth is. Or the ultimate in humblebragging, I guess.


Methodology

Much of Decoherency is the salvaged ruins of my posts on the fallen Google Plus, and while assembling those posts into blog form, I labelled cases containing predictions for the future. I've also included a few other examples from memory and from trawling the blogs the old-fashioned way. I divided these into two topics : science and technology, and politics. One is considerably more objective than the other, so these should provide a complementary approach.

Right then, let's begin.



1) Science and technology

Planet 9




Prediction

Let's start with something completely unambiguous.  On April 8th 2016, not recorded on Decoherency but preserved publically elsewhere in my records, I wrote :
I'm just going to go on the record and state that I think planet "9" is a silly thing, and a year or two from now no-one will care about it any more.
Which if any further clarity is needed, means that I thought the idea of Planet 9 was flawed and that there's nothing out there to find. By sometime in 2018, roughly, sufficient searches should have been performed that the object should either have been found or its existence ruled out.


Reasons

There were two things that bothered me, one scientific and one psychological. The scientific was the poor statistics given to prove the existence object. I think it's a major red flag to say that you can say a tiny sample size (six orbits !) is so completely inconsistent with random chance - it doesn't ring true. What does statistical significance really mean when you've got just one system under consideration anyway ? Not much, I think - we just don't know enough about how planetary systems form anyway.

As pointed out in a comment here, there were other lines of evidence, but as far as I'm aware, these were prompted by the six orbits. If you go looking for consistencies you're likely to find them, so this smacks of bias. To be fair, I mainly read the press releases rather than the papers (and I should know better than to do that...) but there didn't seem to be any alternative possibilities discussed by the claimants at all. "This is consistent with a planet", they said. Sure, but what else is it consistent with ? Did they deliberately try and come up with a strong alternative, or did they just jump straight on the one they liked best ?

The second reason is the show-offy overconfidence on display. Recall Rhys's Law Of Press Releases :
The value of a press release and the probability that the reported discovery is correct is anti-correlated with the grandiosity of the claims.
Why opt for a ninth planet ? "Glory" is the cynical answer, though that doesn't mean there isn't a planet to find. But to make such a claim with such extreme conviction from such a tiny sample size... nope, that's not credible. It's much more likely they wanted to do a press release* for publicity, not because their results deserved publicity. I even witnessed some degree of fanboyism over this, with a few people rather rudely defending their favourite experts against my own skepticism. And that's just silly. Where the hell are my groupies, dammit ?

*There's a lot of caveats to RLOPR. Obviously a reputable scientist with a proven track record is more reliable than Twitter user sheepfondler69**, but let's not go there today.
** Not a real example, hopefully.


Verdict

Total vindication, I think. There's been bugger all the press about this for the last two years, so my timetable was roughly correct. The problem is we still can't definitively rule it out, and it's very hard indeed to prove a negative : according to some estimates, it might be 10-15 years before we can say for sure, though others seem more optimistic. So yeah, I could still be wrong, but I haven't changed my opinion on this one. If I am wrong, well, having another planet in the Solar System would be pretty cool. I propose we call it Pluto.



The EM drive




Prediction 

Another totally unambiguous example, this time from 22nd November 2016 :
I'm just going publically on the record to state that the EM drive does not work and it will go the way of cold fusion and all the other pseudoscientific claims before it. That is all. Have a nice day.
And in a comment from the same thread :
It's claiming to violate a result so well-established it would be basically magic. Of course they can and should test it, just in case, but my prediction is that it's absolute bupkis. Best case : it's some super-weird quantum effect that can't be scaled up. Infinitely more likely : it's a measurement error.
I don't think I can state things any clearer than that.


Reasons

I know even less about the EM drive than I do about planetary orbital dynamics, but the whole idea contradicts physical findings that have been established for several centuries. Given all the past attempts to build perpetual motion machines and the like, history shows that such cases are almost invariably mistaken.

While claims of a ninth planet are at least entirely scientifically reputable, the EM drive is not - but my objections are similar (and the sensationalism is equally dangerous). They claim a truly extraordinary result based on observations of low significance at the very margins of observational limits, just like goodness-knows how many people before them. But scientific consensus is established precisely by this sort of testing and examination. If a test were to disprove it, the consensus would shift (not immediately, but it would). So by its very definition, the consensus view is hard to disprove, and the more tests it survives, the less likely any marginal result is to overturn it and the more probable such results are due to errors.


Verdict

Total vindication again. Here I'm at a serious risk of committing a kind of straw man fallacy - if I were to continuously go around saying, "the world won't turn into a sponge in the next five minutes", pretty quickly I'd objectively be a supremely accurate prophet. Likewise if one goes around only saying that low S/N claims about ways to violate feckin' basic physics will eventually be disproved, one could be unjustly hailed as a genius.

Oddly, a handful of normally intelligent, skeptical people did find the EM drive plausible. I've no idea why, they witnessed all the other hoo-hah about FTL neutrinos and BICEP2 and so forth, so you think they'd be more suspicious. Anyway, there doesn't seem any need to dwell on it. It was worth testing, because I think even really silly ideas should be tested as much as resources permit - just in case - but ultimately there's nothing to learn here.

Of course, there's a risk of circular reasoning : the consensus is more likely to be correct so anything that goes against it is not likely to be true. But the use of "likely" is critical. This is not at all the same as saying, "the consensus is a fact so anything contrary is wrong". So here's a bunch of examples form history of how the consensus shifted - generally, it shifts pretty rapidly when sufficient evidence is provided. Is it perfect ? No, but it works.


The Hyperloop




Prediction

As I wrote back in July 2017 :
I doubt very much this will ever become a thing, but you never know.
Somewhat expanded a year later :
I remain skeptical that this can be constructed on a large scale in the real world in an economically sensible way.
Basically I don't deny that the hyperloop may be technologically feasible, but I do deny that it will become a practical reality. More likely it will prove too expensive to construct - substantially more expensive than its advocates claim.


Reasons

I set these down in August 2017 :
Even if the cost per mile could be made cheaper than railways (and I don't see how it can), the infrastructure costs will still be enormous. If they weren't, HS2 wouldn't be costing us >£50 billion. I suppose it may work in a few cases where's there's nothing but flat, hard wasteland between widely separated cities, but everywhere else ? Nah.
Although there are some excellent comments made on that thread, I generally stand by this (conceding that there are indeed large areas flat hard wasteland in some parts of the world). Developing a mass transit system is going to be hugely expensive. How could a giant cross-continental vacuum tube shoot people around at substantially lower prices than railways, given that rail is an incredibly established technology and the infrastructure requirements are similar ? I don't see it. To clarify my prediction though, I say there won't be more than a handful of major hyperloop routes operating commercially within the next decade, and possibly none at all. I do not mean to say it's utterly impossible, just difficult.

With hindsight I may also have been utilising an airship/fusion bias. Fusion has been twenty years away for much longer than twenty years; airships are definitely coming back this year for reals according to innumerable popular science outlets. Big, grandiose projects fail more often than not, so betting that any new project will fail is the safer bet. Especially if the press go into orgasms about it.


Verdict

So far so good, but it's too early to tell for sure. Reviewing the Wikipedia entry, it looks as though lots of people are still interested and working on this, but there seems to have been precious little hardware testing in recent years. No-one's got a test track going at anywhere near the claimed full speeds, and the economic prospects are at best divided. Although there are some excellent comments preserved in this thread, I stand by my original objections.


SpinLaunch




Prediction

Lordy. Apparently some weird startup company is trying to launch things into space by means of a giant sling. The title of the original post says it all : "space catapults are not a thing and never will be". I was of half a mind that the company might be an outright scam.


Reasons

Again a bias against the success of grand revolutionary projects played its part. Is this fair or is it a sort of techno-racism ? I don't know. And there was an additional "that's just silly" factor, because spinning things round so fast they shoot off into space is, well, a bit daft. Using a CGI image of the factory doesn't help either.

More rational objections : its claimed speed is nowhere near enough to reach orbit, spinning in a vacuum to reduce friction will cause the released projectile to slam hard into the atmosphere when it exits the acceleration chamber, and you have to design a rocket (and satellite payloads) capable of functioning under extreme accelerations (based on a Wired illustration, this would be a bone-pulping three thousand g !). Maybe you could use it to launch bulk materials, but surely not any kind of delicate equipment - and since it seems geared towards small satellites, you're gonna have to have delicate equipment on board.


Verdict

Two years later they're still bringing in punters, raising tens of millions of dollars in funding. They were supposed to have tried a flight test last year, but didn't (though every space development company misses deadlines). It's too early to tell, but nothing has happened that's changed my mind in any way. I think it's stupid.


Honourable mention : VIRGOHI21




Prediction

I never really made any prediction here, which is why this is only an honourable mention. I cannot find my exact words, but I was skeptical of the most popular method used to explain the weird object VIRGOHI21 (the nearest thing I can find to the full description I wrote is here). In brief, I thought that two galaxies experiencing a close, fast interaction would not be able to produce an object like the one claimed.


Reasons

Unlike the other cases, this one is well within my field of expertise. It seemed to me that the earlier model had been very carefully tweaked to produce the best possible outcome but still wasn't convincing. Their main galaxy was too low in mass but too rich in gas, with the gas extended too far, the gas physics too simple. Their final result didn't reproduce the most intriguing and important feature of the system and I thought the scaling on their figure was misleading.


Verdict

After trying this for myself with simulations that corrected the apparent deficiencies, but also used many more galaxies, I have to admit that the central claim is likely correct. That is, an object such as VIRGOHI21 can indeed be produced by tidal encounters between galaxies. I still dispute whether this explanation is the correct one, however, as consistency isn't evidence : other explanations could be equally consistent and have not been tested. I think my doubts were reasonable and warranted, and the simulations worth running, but the main conclusion - the important bit - from the earlier paper is correct. So everyone's a winner here, but I was wrong in my central doubt.


Conclusions to part one

It seems apparent that 1) I'm wary of sensationalism; 2) I don't trust revolutionary claims; and 3) I've chosen easy pickings. If I really want to question my scientific world view, in future I should make predictions for more difficult cases that require greater understanding. Otherwise there's not much to learn here - I don't have an ideological stake in a big spinning vacuum tube or a missing planet. If I'm wrong about them, I can all too easily attribute this to a poor knowledge of business or orbital dynamics, which don't come with any moral or political elements attached. There's not much "belief" at work here, just knowledge with gaps. Filling in those gaps would make me less ignorant without having to rebuild any supporting structures. In no way is my identity attached to any of these ideas.

Or to put it another way, I don't wake up every morning and loudly proclaim,  "ahhahah, I bet the Planet 9 people feel really stupid today !". Until I started writing this post, I'd basically forgotten about it. Casual criticism on the internet shouldn't be taken for an obsessive vendetta. I never reached the stage of caring very much about any of these examples - sure, VIRGOHI21 is damned interesting, but that's not the same as being personal. Criticism of findings is, sometimes, genuinely directed only towards the findings and not the scientists who made them. I'm sure the planet 9 bunch are lovely people, though I'm less sure about the EM drive gang or the Spin Launch crew. They might well be idiots. albeit obnoxiously rich idiots in the case of Spin Launch.

Still, surely I shouldn't criticise fields outside my area of expertise ? In general yes, but some fields are cross-discipline. If I saw a study claiming that lawnmowers are the sole drivers of global warming based on dodgy statistics, I'd absolutely be entitled to debunk that. And criticism is not all the same as censorship : if I say, "I'm not an expert in this but I don't believe it, here's why", I'm emphatically not automatically calling for that research to be ended (I even said as much in the EM drive case). Ideologically, if not practically, I'd prefer to test everything, especially the things I disagree with because those are the most interesting. If I ever do call for research to be ended, I'd like to think I'd have some very good reason for that.

In short, when I say, "this is stupid", what I really mean isn't, "stop giving these people money", it's, "everybody PLEASE shut up about this and talk about something better now". Super-saturation with the latest improbable finding is the nerd equivalent of magazines full of discussions about vegan yogurt smoothies and celebrity hairstyles.

More interestingly, there's a conflict here as to whether I want this technological claims to be correct or not. I'll cheerfully admit to wanting to be able to say, "I told you so !", but at the same time, it would be genuinely cool to have a reactionless drive or a giant space catapult. Even without joining any fan clubs, the effect of bias isn't simple.

Let's move on to the more complicated cases where I do have a more emotive stake in the outcome.



2) Politics

Theresa May's fate



Prediction

As Stephen Pinker helpfully reminds us, "things that can't go on forever can go on for much longer than you think". Boy oh boy was this true of the sad tale that is Theresa May's government. Back in April 2017 I wrote :
For Theresa May it's do-or-die at this point : either secure a "mandate" from the populace or accept defeat and a potential change of course.
And in June 2017 I wrote :
She'll try to form a minority government with the DUP, and it will work for a little while but not for long...  Sooner or later - probably sooner - she'll break.
And then in October, when rumours were flying of a no confidence vote :
I suspect May will limp on for a little while, but this is probably the beginning of the end.
Then in July 2018, after a series of high-profile cabinet resignations :
I don't see even May managing to survive this one. She can't even look her colleagues in the eye. 
 But by December, just before a confidence vote finally happened :
I bet she wins.

Verdict

Wow, did I get things wrong. For more than a year I was predicting May's imminent downfall, but she didn't actually break until July 2019 -  two years after my initial prediction, whereas I was typically thinking she'd be gone within maybe six months. Only by the time the confidence vote happened did I realise that her track record indicated a pug-like ignorance of reality. May did eventually resign and her government did eventually break, but the point of the prediction was that it would happen soon - which it obviously didn't. As for the "accept defeat or change course", well, that one's somewhat a grey area : she won a majority, but substantially smaller than previously, and utterly refused to change course.


Reasons

I completely misjudged her character. I genuinely believed it when I said :
May does not do well under pressure; she called the election out of a peculiar sort of desperate opportunism, and we've seen her increasingly degenerate into robotic performances that make little or no sense.
Thing is, she doesn't do well under pressure. She doesn't do well anyway. But either she was totally unaware of how bad the situation was or just didn't care (or, perhaps, was genuinely trying to mend fences with unpleasant colleagues but going about it very very badly). I thought her robotic TV persona was just a way of masking her feelings and avoid having to deal with anything unpleasant, but, in fact, she might actually just be like that.

More charitably to May, I'd also misread the wider political context. For all her inefficacy, May was right about one thing : the facts. She had them on her side a lot longer than her detractors gave her credit for. Oh, she misused them with consummate skill, but it was really true that her stepping down earlier wouldn't have changed anything for the Tories and might well have left them with someone worse. I think I do have to accept a strong ideological preference on my part - a desperate hope that she'd leave and somehow things would get better.

Overall, 1 out of 4 (ignoring the ambiguous one about a change of course) is dismal. My assessment of the situation was biased and wrong.


Labour's leadership crisis



Prediction

Back in October 2016, Labour prepared to choose between incumbent Jeremy Corbyn and challenger Owen Smith. I wrote :
This is quite likely to result in the next major crisis in British politics. Oh, yippee.
I didn't really elaborate , but what I vaguely recall thinking was that either way we'd get factionalism that would split the party. Corbynites loved Corbyn and hated everyone else in the party, and vice-versa. A recipe for disaster for sure ! I predicted a Labour party split at least as early as 2017, in the event that Corbyn continued to be Corbyn :
Best case long term realistic solution ? Labour have to split.
Once they realise that and find that they're sliding back into the doldrums, all the old animosities will re-surface and they'll split.
But I'm pretty sure I thought that was a distinct possibility quite a lot earlier, though I can't find a record to back that up. Anyway, I thought that there was a very real possibility the Labour party would have been in such dire straights as to have or be contemplating a split by about now.


Verdict

Obviously, that hasn't happened, so again I was totally wrong. After leading Labour to disaster, Corbyn finally quit - thus rendering a split pointless. All the more so as - for now - the Corbynite movement is pacified with a leader acceptable to all divisions of the party.


Reasons

Corbyn seemed to have even more resilience to the facts than May. 90% of your own MPs want you gone and you think you can just carry on ? Foolish beyond belief. But I was mistaken about Corbyn's lust for power, which was not quite as bad as I feared *. I was also wrong in suspecting that the Labour party would refuse to grin and bear it after Smith's leadership bid failed (though if recent reports are correct, they tried a much stealthier tactic than I would have guessed, meaning I was still wrong but in a different way). Electoral disaster did happen under Corbyn, as I thought it would, but I failed to see that the party would then finally see sense and decide by some considerable margin to distance itself from the hard left.

*But only by a little, since the guy hung around for bloody ages after losing the 2019 election so badly.

So here I misjudged both the individual character of Corbyn and the group behaviour of the Labour party (the latter twice, for their response to each election). This is pretty bad, but not quite as bad as it appears. Corbyn was just a bit less of a shit than I thought, whereas the Labour party were either more sneaky and/or less principled, or possibly more weary, than I supposed. I think these mistakes arose from relatively minor misjudgements that had big consequences.


The Liberal Democrat resurgence(s)



Prediction

An interesting mix here. In April 2017 I was hopeful and maybe even expectant of a significant Lib Dem breakthrough :
The Lib Dems recently won some spectacular victories in by-elections... Anecdotally, I know too many once-devoted Labour supporters (both young and old) who are literally disgusted with Corbyn to take any claims of a shock Labour win seriously.
.... the "safe seats" idea appears to be passing, given both the previous general election and recent by-election results. Cardiff voted for Remain, which makes a Lib Dem surge here not so implausible.
Of course I didn't expect them to win or even come close :
I accept that we won't get a shock Labour or Lib Dem win, but would a Tory loss be so unexpected ?...  I do think there's a chance of an upset. 
And while we're here, as to the Tories :
The Tory minority is tiny. It's far less implausible to suggest that it might be reduced to nothing and the government replaced with a coalition of the left.
Whereas in November 2019, despite the Lib Dems having secured more seats via defections and performed extremely well in European elections :
I'd rather have the Lib Dems over Labour, but in Cardiff North that doesn't look like a realistic choice.
And I finally decided to vote Labour, despite despising Corbyn, on the grounds of tactical voting (which I also disapprove of). Oh how principled I am !


Verdict

There was no Lib Dem resurgence in 2017, so I was wrong there. The Tories did, however, reduce their majority substantially, though not completely, so I give myself partial credit on that score (and Labour lost, so I was right about that too). In fact the government even fell to an impotent minority, but only after months of by-elections and defections rather than due to the election itself. And I was right - having learned from bitter experience - that a Lib Dem choice just isn't realistic in Cardiff North, so choosing Labour was at least better than throwing my vote away to the Lib Dems. So full spectrum here : totally wrong about the Lid Dem resurgence, partially right about the Tory losses, right about Labour not winning, and right about the local Lib Dem failures.


Reasons

With hindsight, I cannot honestly tell you if I was merely hopeful or truly expectant that the Lib Dems would perform much better in 2017. I think I may not have even known myself. On the other hand, I was quite worried by voting Labour in 2019, thinking that I might be betting on the wrong horse - but a doubt in the back of the mind is not the same as an actual vote cast. I largely went with gut instinct in 2019 rather than any careful thinking.

Why wasn't there a big Lib Dem resurgence in 2017 ? I don't know. In 2019, overall I didn't believe they had much of a chance, having failed to convince me of their credibility and Labour - just barely - having done enough to satisfy me regarding the main issue of the day. But I have no idea how typical this way of thinking was. The first-past-the-post system (more on that below) means we should be extremely cautious about analysing overall results, so this remains confusing.



The 2019 election




Prediction

Here I have to say that I was outright silly, and fell into all the same traps as all the popular left-wing commentators. In a deliberately rhetorical piece in September 2019 I wrote about Boris Johnson :
Expelling MPs en masse is unlikely to be something the Tory party machinery is going to reward come campaigning season.
He can't even keep his family loyal... His singular approach of bully and bluster is fine for preaching for the choir - that's why he won the PM election with the Tory faithful - but useless for winning hearts and minds, which is vital for winning back Tory control of the House.
 Though I did temper this with :
While things don't look good for Boris come election time, we should remember that he's largely untested in such a campaign - even fools have a few strengths. Given the first past the post electoral system, it is still credible that, if the pro-Remain parties don't tread carefully, we could end up with a Tory majority. 
I was quite pleased with the rhetoric in that post, but that does make it harder to get at what I really meant. I was, I think, genuinely quite confident that Boris Johnson could lose the election. I added in a note of caution only to say that we shouldn't take anything as a foregone conclusion, but this doesn't negate my totally mistaken prediction.

In another deliberately rhetorical post a few days later I concluded somewhat more successfully :
It is possible, however unlikely, that an election could result in No Deal or further prevarication. But it is also possible, and considerably more probable, that it could end the whole sorry affair.

Verdict

Even with my caveats about success not being certain, it's pretty clear that I actually thought Boris could very well lose. He didn't. But there is an important factor to remember here, one which, ironically, I've been at pains to examine before : the first past the post system. I mentioned this in the predictive posts, but badly failed to understand how it could play out. As far back as May 2015 I noted that :
The SNP's sweeping victory disproves any idea that a different system is necessary for the success of smaller parties.
Though that's a bit out of context as the original post was fairly nuanced. But the point is that the Tories did not win the popular vote; in fact, 52% of the vote went to pro-Remain parties. So even though I certainly underestimated Tory popularity, I also had a valid point. I was however utterly wrong in thinking that the Tory party might not support its own leader. This isn't quite as stupid a suggestion as it might sound, given recent reports that the Labour party may have tried exactly this during the 2017 election.


Reasons

The FPTP system makes the thing very difficult to analyse without doing detailed breakdowns on the numbers, which I'm not prepared to do here for the sake of a single paragraph. One thing I can state with confidence is that I overestimated Jeremy Corbyn's campaign ability, which this time wasn't as successful at clawing Labour back from abysmal polling figures as it had been in 2017 - the public, I think, were just fed up with him. Conversely, I probably underestimated Boris Johnson's popularity too, as well as misjudging the abilities of each party to work with the FPTP system (this, though, has led to a substantial shift in my opinions on proportional representation, discussed here).

I also think I may have simply been too close to this. With a vested personal interest, it's easy to get caught up in the daily sensationalism whereby every minor error of a politician is depicted as grounds to break their legs and throw them to the wolves - particularly if you happen to have a really profound dislike for them. And more cynically, it's arguable that the public just aren't interested in this stuff. Whereas an enthusiast of political theory will see Boris Johnson hiding in a meat locker (or the now-President of the Ukraine saying nothing much at all throughout his whole campaign) as a profoundly undemocratic act of avoiding scrutiny, the general public just don't care.


Honourable mention : Jeremy Corbyn

Not really a prediction but a simple case of changing my mind. Since this is described at length here, I'll just summarise. Initially I thought Jeremy Corbyn was a hugely underrated, charismatic, game-changing individual who was treated incredibly badly by the popular press. Now I believe he's a feckless, smug, patronising, authoritarian.... well, basically a horrible little man who I wish would just sod off. However, I did follow my own advice, which was to give him a year and see how he got on. The answer was, "terribly", and so although my initial assessment was way, way off, at least I changed my mind about him much sooner than most of the left-wing tabloids. Even now, almost unbelievably, there are still commentators who think he should have magically won and the Labour party should stand by him for some strange reason. It's pretty clear that my political prediction skills are frankly shite, but hey, at least I'm not that bad.


Conclusions to part two

I am way less good at remaining objective in the face of ideological preferences that I would have expected. I'm not wrong about everything, but I'm wrong a lot. Granted, if you go back and look at some of the cited posts in their entirety, you'll find that there are caveats - I don't often pronounce things as definitive either way. But the fact is that even when I'm only confident enough to make a "most likely" prediction, more often than not, I'm still wrong.

By my count, politically I'm wrong about twice as often as I am right. And what's worse, my correct predictions were all really easy ones. I can even add a few more failures from memory : at one point I was quite convinced a second referendum was inevitable, and I thought Corbyn's decision to admit he wasn't going to campaign in such a referendum would be something that would appeal to voters by virtue of its honesty; I also though there was a real possibility of Donald Trump starting another war; I didn't think Boris Johnson had any chance of getting a Brexit deal. So yeah, I'm bad at this.

Ideology has certainly played its part in blinding me to the obvious here. Even though my own moral viewpoints were not directly at stake (for example. that the Tories were re-elected doesn't vindicate austerity, nor does thinking Jeremy Corbyn is a total arse-monkey make me question renationalisation) my mere preference was enough to cloud my judgement. I didn't pin my reputation or identity on any of these predictions, but I still got them wrong.

There's also an interesting but complicated contrast here with the technology predictions. For those, I predicted failures but would have preferred it if they'd succeeded, whereas for politics I predicted failures and wanted failures. If preferences do influence beliefs and predictions, clearly this isn't as clear-cut as I first thought. The desire to be right for one's own sake (to blow a raspberry at people and say, "I told you so !") is not the same as the actual outcome wanted. Perhaps that's a factor I should consider more carefully in future predictions.

In fairness, sometimes I've been nearly right, but small errors can lead to big consequences. And some suggestions which may seem obviously wrong are not as daft as they may first appear. Still, undeniably some things I've said have been totally stupid. On the other hand, try and find me someone from the Independent or the Guardian (never mind the Mail) doing a similar exercise. Go on, I'll wait.



Summary


Ahh, the thankless life of the internet commentator.
I decided the easiest way to summarise all this and record things in the future was by means of a spreadsheet. I'll update this with future predictions too.

It's pretty safe to say I do better at science and technology predictions than I do at politics. Maybe I didn't put much basic ideology on stake in the political examples here, but I definitely wanted certain things to happen and certain people to be right or wrong. In contrast, in some ways I don't give a monkey's whether a whirly space vacuum tube thingy works or not. I mean, besides saying, "wow, that's cool", what's in it for me ? Nothing*. And we can easily and objectively test whether the whirly space thingy works or not, whereas judging someone's underlying motivation is always going to be at least a little bit subjective.

* Besides a massive nerd boner. That is, until space travel becomes so cheap as I can afford it myself, I don't have much of a personal stake in it.

If nothing else, I've discovered that it's damn hard to say what my real preferences are : I always want my prediction to be right for its own sake, but that doesn't mean I always want that outcome to actually occur. That'd be like yelling, "I told you so !" as the Titanic hits the iceberg. Maybe in the future I should more explicitly state what it is I'd prefer to happen. While praise and shame are reputed to be major drivers of belief, this exercise has made me wonder just how true that is - and reminded me that bias can act in different ways all at once.


A much simpler factor is that politics is just a lot more complicated : there are many more variables and assumptions I'm not always aware of; things I take for granted or never take the trouble to properly scrutinise. I'm never going to say, "Theresa May will continue to be unpopular, unless she manages to fight off a bear on the steps of Downing Street". Politics is a lot more provisional than technology - it has a lot more unknown unknowns. I have to assume Theresa May won't be attacked by a bear, otherwise I also have to consider the possibility of her being eaten by a lion or falling deeply in love with an inflatable sheep, or tripping and breaking her ankle in a field of wheat. Or something.

One thing that prompted this post (aside from the oodles of spare time) was the observation that professional political commentators seem to be truly adept at tribalistic bullshitting. I even made this lovely little chart, which I can simplify here :

How to very cleverly justify something really stupid. If I were to plot my predictions here, I suspect there'd be some in every quadrant.
Professional commentary on all sides seems to gravitate strongly towards that upper-left quadrant. They're darn good at persuading you of anything they like, but they don't often stop to acknowledge if they were wrong in the past. They're concerned far more with who than what - if someone they dislikes says something they approve of, it's invariably due to an ulterior motive or still inadequate in some way. Hence you get bollocks like holding different people to truly ludicrous double standards and absolutely no inkling of whether an author is truly being fair on just going on an angry rant. And what's particularly horrible, particularly insidious about the whole thing, is that even being aware of this doesn't automatically make you more skeptical or less vulnerable to letting confirmation bias get the better of you. I still head towards opinion piece headlines I approve of.

Recently I tried out a quite nice tool from Clearer Thinking, which is supposed to help you challenge your own beliefs. You give it a statement and then it guides you through the process of critiquing it. At the end, all I found was that my new, revised statement had just been made a little more watertight, proofing it against unforeseen circumstances (and/or enraged bears) but not changing the substance of it one bit. Bullshitting, it's worth acknowledging, is a perilously easy trap to fall into.

Knowing how wrong I've been, should I now re-evaluate my beliefs ? Well, clearly I'm not a great judge of character, so at the very least I should be more cautious in giving the latest political leader my support or disapproval. On the other hand, I will stand up and say I'm ready to change my mind about them in according to the evidence - at least, at lot more so than the gutter press. And while I'm actively annoyed by science and technology sensationalism, I'm much more vulnerable to the political kind, so I should try harder to keep it at arm's length. Hopefully being able to look back on my own record here will help remind me of that. Also, I've overestimated my ability to judge the opinion of the public and political parties; I've made several predictions when I had no access to the necessary inside information.

Finally, I haven't really made any predictions that directly flow from my deeper ideological beliefs, e.g. whether higher taxes would be a good thing or not, whether we should relax austerity or not - I've been concerned too much with individuals and day-to-day details. This means I've kept those beliefs shielded. I think they're backed up by existing evidence, but if I really understand them properly, I ought to be able to predict their effects in the future too.


One last point is that even an opinion formed in the upper right panel of the chart is not necessarily right or wrong, and likewise for the others too. Sheer luck, caused by all those nasty complications that is the mess known as "real life", means that even the stupidest person can sometimes outfox the cleverest. But generally, surely it's better to be rational than wholly irrational. So if we did get some analysis of how often the major pundits get things right or wrong, it might indeed be a good guide as to who's worth listening to and who's just blowing their own trumpet. Perhaps we could make an app to make this easier, or even teach it in schools so people will grow up without being afraid of being wrong. On the other hand it might just encourage even more bullshitting by making them too afraid to say anything of real substance at all, leading to garbage language that's utterly incomprehensible. That's not a prediction I'm willing to make.

Even with all this uncertainty, I'm willing to offer some take-home points :
  • Bias is complicated. We can simultaneously desire to be right and wrong, so how this influences a prediction isn't always straightforward.
  • We can be wrong due to simple ignorance, implicit assumptions, or fundamental mistakes. Not every wrong result means we should change our whole world view.
  • We can get things right for the same reason : essentially, sheer luck.
  • If we want to use predictions to test our ideologies, we have to expose those ideologies to direct testing rather than just testing individual characters and day-to-day occurrences. We need to specify clearly and ahead of time what we think will happen and why, and how different situations and conditions might play out. Otherwise we have no standards against which to judge ourselves, allowing us to get away with anything.
  • Being wrong isn't always fun, but it's usually a learning experience.

2 comments:

  1. This comment is left here to assure Dr Rhysy that at least one ex-G+ fanboy is still reading his blog.

    ReplyDelete
    Replies
    1. This comment is left here to express Dr Rhysy's delight and gratitutde at receipt of this information.

      Delete