Changing my mind

[liv]

I aspire to be the kind of person who thinks for myself and most importantly changes my views when I learn new information. And that means I spend some amount of time worrying about whether I'm actually living up to that. There are lots of pressures pushing people, including me, towards opinions that are not based on evidence and convincing reasoning. The biggest one is probably conformity, I do have a strong drive to hold opinions that match the accepted ideas in my social circle. And that's a bit self-reinforcing, because I'm likely to be friends with people I broadly agree with and then when I hear arguments from people I already like I'm more likely to be swayed by them. Plus there's all the usual biases, I'm more skeptical of evidence for ideas I disagree with and relatively receptive to arguments that support views I already hold. And like many people I find changing my mind a bit scary because it implies repudiating the person I once was who held the now discarded view.

Part of the reason this came to mind lately was that some people were passing round a link to Scott Alexander's persuasion experiment. I find Scott Alexander very annoying for a lot of reasons, particularly his whining about how a feminist was mean to him one time and therefore feminism and all of social justice are an evil cult. But he does sometimes have some useful things to say and one thing I do admire about him is that he's quite committed to being precisely the sort of person who does update his ideas based on rationally examined evidence. And that's not my major goal in life but it is something I care about.

So anyway, I tried reading a lightly randomized essay aimed at convincing me to be scared of what the Rationality crowd call AI threat. And I was not at all persuaded, because although I did my best to be open-minded about the idea that I should be scared of the development of an all powerful, self-improving AI bent on the destruction of humanity, I find the concept so obviously ludicrous that no amount of writing in a pseudo-academic style with footnoted citations was going to shift my position. Plus I'm prejudiced against Alexander (as far as I know he didn't write the essay, but he promoted it) and fear of a powerful, destructive AI monster is a trope that I already associate with my negative feelings towards the Less Wrong / New Atheist / Rationality set. I think it's exactly an example of the kind of errors you make when you over-estimate your own intelligence and try to understand the world just by thinking really hard and without actually engaging with existing scholarship.

This did lead to a bit of self-doubt: am I just losing my ability to change my mind? Should I in fact be worrying about AI threat and I'm just blocking myself from doing so because I don't want to challenge my existing views and biases? (Actually I think the whole debate is directly analogous to Pascal's Wager: the fact that a clever person can invent an outcome that would basically be infinitely bad doesn't mean either that that outcome is likely, or that we should spend maximal amounts of energy trying to avoid that bad outcome.)

Then I did change a previously held belief, because people on my Tumblr dashboard were talking about this meta-analysis of studies about the effects of punishing children by spanking. Unfortunately the study is paywalled and I can't quickly find the Tumblr post where someone had put up a pirate version of the full paper. But basically Gershoff convinced me that there is a clear evidence base against spanking, given that she found lots of studies with small but reproducible negative long-term outcomes and essentially no studies with statistically meaningful benefits. But I don't really feel proud of myself for changing my view here, because for one thing basically all my friends are against any kind of corporal punishment so I'm just removing what was previously an outlier belief and replacing it with the group consensus. And for another it's a completely abstract question since I don't have or plan to have children so it makes no difference how I regard the arguments about childrearing practices. Besides, siderea and rmc28 had already half convinced me when I mentioned the issue before. So it was really quite easy for me to shut up the critical voice in my head pointing out that Gershoff has based her whole academic career around campaigning against corporal punishment and psychology has a replication crisis and a meta-analysis can be cherry-picked... You know what, self, that's as good a meta-analysis as plenty that you'd unquestioningly accept if they didn't require you to change your opinion, so there's no need to be overly picky about this one.

So those are the two extremes. I'm unpersuaded by an article espousing a view I think is not just wrong, but ridiculous, and more so because it's written in a style and associated with a group I disapprove of. I'm persuaded by a peer-reviewed meta-analysis to change a view I was only mildly committed to anyway to one which is more aligned with my social group. What I'd like from my readers, if you'd like to play along, is for you to persuade me of some new ideas. Please send me links to arguments you find persuasive on issues you expect me to disagree with. (I'm also quite interested to discover what you think I might find objectionable; I think I've been pretty open about my opinions here over the years, but of course everybody will have their own impressions and assumptions about me.)

I've turned off screening for anon comments, so if you think your views might be met with social opprobium please feel free to offer arguments without saying who you are.

Comments

[altamira16]

I live in the US, and one of the hot topics of debate is whether genetically modified organisms should be labeled or not. The arguments for this is that we should have transparency and know what is in our food. The arguments against this are that labelling can really drive up food costs for people who have to watch their food costs. If everyone decides that GMOs are bad, products with GMOs become less readily available because people are no longer buying them. Where I live organic produce can cost two or more times what conventional produce cost. Here is an article about an anti-GMO group sending a freedom of information request to a pro-GMO scientist. Here is an article that is critical of Food Babe.

[liv]

Ooh, that's a really interesting one, thank you! As a genetic researcher I'm completely pro GMO, and as a capitalist I'm generally in favour of labelling and consumer information, so I am not sure what my view actually is about GMO labelling. The Food Babe article generally agrees with what I already believe, that a lot of the anti-GMO movement is just anti-scientific crackery and I have no time for it.

But the FOI stuff, I'm really not sure. I mean, in general I am in favour of members of the public being able to see correspondence between scientists and industry, because there definitely have been problems in the past with conflicts of interest. Like the anti-fat, pro-sugar propaganda stuff that's been in the news recently. Or medical researchers being all but bribed by drug companies to report favourably on their medicines. My first reaction is that those particular FOI requests sound basically vexatious, and I can see the point the article makes about the risk of intimidating scientists.

[rysmiel]

If people label things as not containing GMOs, I can then boycott said people as being anti-scientific.

[seekingferret]

I also tried the persuasion experiment and likewise was not persuaded that arguments about AI Existential Risk are anything other than bad argument by analogy bolstered by incorrect ideas about how intelligence works. I think you're also right that the analogy of Pascal's Wager is relevant, at least that this is a test of our probabilistic intuition about the future- a lot of this Existential Risk stuff is only scary because the very slimly possible worst case scenario is very bad.

[liv]

To be fair I did kind of bias my readers against the AI persuasion since I presented it in such a negative light. And I think you're right that failed intuition about risk and probability is a key flaw in the reasoning, and a lack of sensible consideration of what machine "intelligence" would be like is another issue. (I give Pascal himself some credit for basically inventing formal probability and little blame for not having a modern understanding of infinity, mind you.)

[woodpijn.livejournal.com]

I think you're painting quite a distorted caricature of Scott's position on feminism.

Re AI risk, Scott and other rationalists have successfully persuaded me that if an AI had human-level intelligence there would likely be bad unintended consequences; but I'm not signed up to the AI risk movement because I don't think we're likely to be able to build such an AI. But I don't know whether that's a rational scepticism on my part or just a knee-jerk "humans are special" reaction.

[seekingferret]

What does 'human-level intelligence' mean to you? Nobody's given me a clear explanation of what that would look like in a computer, they seem to just pass it off as a term that anyone would intuitively understand.

Scott's experiment used the heuristic for 'super-intelligence' of 'far more capable at nearly every intellectual task than any human', and I have a lot of problems with that heuristic, but trying to down-translate it to human level intelligence 'as capable at all intellectual tasks as a human' seems completely unintelligible to me because that could mean basically an infinite number of different things.

[liv]

You're right, I am being entirely uncharitable in my summary of Scott's criticisms of feminism. But I've read several of his long essays about why feminists and SJ activists are some combination of a religion and a censorship board and opposed to evidence and rational thought, and I am almost completely unsympathetic to that view.

I think bad unintended consequences to very powerful AIs is reasonably likely, and I think it's reasonably likely that AIs will continue to get more powerful. But I think there's no basis at all for the leap from "something bad may happen" to "AIs will wipe out all of humanity". Further, we're at permanent 400 ppm atmospheric CO₂ now, and we're really rapidly running out of antibiotics, and the US is at about evens chance of electing a president with no self-control and giving him access to enough nuclear weapons to kill every living thing on the surface of the planet... and the AI risk people are arguing that we should put all our effort into preventing the development of some totally hypothetical AI that's intelligent enough to defeat all possible defences against it but not intelligent enough to work out that killing all humans may not actually serve its goals. It's not that I think human intelligence is particularly special compared to computers, it's that we have much more pressing and realistic problems.

[forestofglory]

I am beginning to think that there are some issues I am willing to change my mind about and others I am not.

Let take two issues that I am going to vote on soon. (I'm in California where we have a lot of referendums).

The first issue is legalization of marijuana. I planning to vote in favor because I think it will do more good than harm. However I am open to evidence otherwise. In fact did so research on the environmental impact of legalization before deciding to vote in favor.

The second issue is the death penalty. I strongly oppose it and feel it is morally wrong for the state to kill people. There's a poorly crafted measure on the ballot that eliminates the death penalty. I'm planning to vote for it dispite it's problems because I oppose the death penalty so strongly. (It is possible someone could change my mind on the measure, but not on the death penalty.)

Sometimes I just have really strong feeling about issues that aren't very rational.

[liv]

That's a really good point, there are definitely some issues I'm not prepared to change my mind about. I don't think there is any possible evidence that could convince me I should be intentionally racist, or that human industrial activity doesn't affect the global climate. And I'm not really interested in changing my religion and some of my moral axioms, because those are too much part of who I am and not really based on an opinion as such.

Really interesting point about the environmental impact of legal marijuana! I am basically in favour of legalizing / decriminalizing drugs, but I don't think this country is likely to be a major producer of hemp and marijuana any time soon, we have the wrong climate. But I should perhaps think more about the effects on the ecology and economy of source countries of increasing international trade in drugs.

I would generally rather have bad laws against the death penalty than have the state executing people, so I can definitely see where you're coming from with that ballot measure. I think I slightly believe that it was right for Israel to execute Eichmann and for Iraq to execute Saddam Hussein, though, but that's not a strong enough belief that I'd actively campaign in favour of the death penalty.

[lilacsigil]

I first ran into the AI threat idea as Roko's Basilisk which is something closer to a religion than an argument, which has made me incapable of considering it seriously as an argument rather than as a thought experiment.

Another possibility about not changing your mind as much might be that as you grow older you have been exposed to vastly more different opinions and life experiences, and your positions are generally more considered - the burden of proof to change your mind is now higher. I can't think of anything where I've completely changed my mind in the last 20 years or so but I have modified (and usually moderated) my position on many things.

[liv]

Absolutely. My first reaction was that this supposed "AI risk" is just a more sciencey sounding way of talking about Roko's Basilisk, which is a quasi-religion based on a really ill-founded thought experiment. Not tremendously impressed, really.

Yes, I think you're right that my positions are more entrenched because I have actually had more time to think about them, rather than because my brain is getting inflexible with age. I find that a comforting thought, so thank you. There are definitely some topics where I really have already looked at the evidence and I have little interest in rehashing arguments when I've already decided. But I do want to be careful to avoid getting too comfortable in my views because I don't want to be one of those old people who stick to the bigotries that were current when they were younger and can't keep up when society moves on.

Inflexibility of Old Age

[(Anonymous)]

A side issue, but I think the following is an aspect to be considered as another factor making older people more reluctant to change their opinions.

One of the reasons I retired at age 70 is that I realised a tendency to handle my work along familiar lines rather than making a more rigourous analysis of the particular issue.

In many, if not all professions, I feel that this tendency can be a disadvantage of long experience.

Self-awareness of one's limitations can reduce the risk but there is, of course an effective solution - training the next generation. I cannot think of a more efficient mechanism of continuing professional development than training novices. Being obliged to concentrate on first principles has enormous benefits for the trainer.

[(Anonymous)]

Here are the things I can remember changing my mind on as an adult, which I felt strongly about as a teenager. Not quite what you asked for, but might be interesting seeing if you still agree with teenage-me (some of them you obviously don't, but I was interested in making the list!).

I've split them up in seperate comments in case people want to reply to make the threads easier to read

[(Anonymous)]

*) Grammer schools are good -> grammer schools are bad (mostly because 'grammer schools were good for me and I am making policy based on over achiving bright kids who are having a miserable time being bullied' moved to 'grammer schools really screw over the people near the artificial line, are a type of streaming rather than a type of setting, which is bad, are based on one arbitary test at one point in time, which is bad, and have a big pile of evidence that they mostly benefit rich people and aren't all they're cracked up to be for social mobility'. I _think_ this is mostly regression to the social justice mean, although I have a lot of pro grammer school friends.

[(Anonymous)]

*) Trans is just a mental disorder where people believe a Not True Thing, like thinking they are a cat, or Napoleon -> Gender is really really complicated, and clearly can't be defined simply in a lot of cases, so trusting people in their lived experiences (and not thinking my need for oversimplistic boxes is more true than something they have thought hard and intelligently about) seems a better, truer, model. This again is regression to the social justice mean, I think.

[(Anonymous)]

*) Abortion is bad, and the legal limit should be really really early -> Abortion is bad, but should be legal until birth. None of the arguments I heard about this depended at all about how old the foetus was, with the exception of some very ablist ones, and some very judging 'if the mother can't cope enough to organise an abortion in three months, then she deserves an unwanted pregnancy and birth'. I really don't think there is a huge population of mothers out there wanting the horrific trauma of late pregnancy termination, and I think we should be supporting women who want abortions to have them as early as possible, as well as changing as much of the world as we can to reduce the number of those women (better birth control, better benefits system, etc etc), but it should not be against the law to abort at any age. Not sure if that means there should be NHS provision of late stage abortion, but then I don't really believe there's any demand, and if there is demand it's a sign things have gone wrong elsewhere in the system and need fixing. (I think 'abortion should be more legal' is regression to the social justice mean, but I'm not sure I know anyone else who argues it should be legal up to birth)

[(Anonymous)]

*) Borders are just the way the world is and I haven't thought about this very much but we need them -> Everyone should be allowed to cross every border, and live where they want and work where they want.

Probably my most radical belief, brought on by a range of things, including

- a lot of very personal looking at people I care about being completely shafted by our immegration system for really stupid reasons (eg an Indian friend who couldn't come on holiday for two weeks because he wasn't married and so the UK decided he had 'no reason to want to return to India')

- a lot of realising that so much of the anti-open-boarders is not only racist scaremongering, but it's _wrong_ - things have to be really horrific for most people to have any interest at all in leaving their family home and where they grew up, the numbers of people who do this to see the world/ develop careers are a small proportion, and actually make the world a more awesome place by doing it, shengan did not mean every Pole in Poland moved to Germany

- a lot of realising that even if everyone in Outer Foreign did want to move to the UK, which is Not True, and they would immediately pass laws banning morris dancing forever, which is Not True, that doesn't mean we have any moral right to state where people are allowed to go based on the accident of their place of birth. This earth was made a common treasury for everyone to share, and people are going to look back on our borders with the same view we look at the bantustans of South Africa.

[(Anonymous)]

*) 'Fox hunting is definitely bad and should be banned, cruelty for fun is awful' -> 'God, we do a lot of cruel things to animals in the name of pleasure, or at least non essentials, the meat industry being the big one, but also leather for Cool Clothes, and just the fact that we hit a lot of animals with cars because we think we have the right to drive everywhere, fox hunting is clearly not about cruelty to animals it's about class, and judging other people (my pleasure from this burger is a natural pleasure, yours from tearing up an animal is a Bad Pleasure) I would probably rather be a fox who lived in the wild and then died in a hunt than a battery chicken, maybe people should decide for themselves if they want to hunt foxes, I sort of hope most of them will decide they don't, but I'm not a vegetarian yet.' (This does not appear to be regression to the social justice mean, I don't think I know anyone else who thinks we should relegalise fox hunting)

[(Anonymous)]

Going anon because I fear social oppobrium, I basically agree with this essay on feminism and and transgenderism - http://www.nationofchange.org/2015/01/08/feminism-unheeded/

In particular, this bit:

"If one understands gender categories (man and woman) as being primarily socially constructed, then trans ideology actually strengthens patriarchy’s gender norms by suggesting that to express fully the traits traditionally assigned to the other gender, a person must switch to inhabit that gender category. For years, radical feminists have argued that to resist patriarchy’s rigid, repressive and reactionary gender norms, we should fight not for the right to change gender categories within patriarchy but to dismantle the system of gendered inequality."

I can completely understand why individual people opt for switching gender category rather than choosing to fight to dismantle the system of gendered inequality - and I try to be as supportive as possible to my transgender friends - but, politically/philosophically, I just don't think transgenderism and feminism are compatible. Of course, you can be a feminist and make unfeminist choices because they are what you need to do to be happy/live your life - I make plenty of those myself - and I think we're all people trying to do what we can in an imperfect world, but I don't subscribe to the "feminism is about choice/all choices women make are fenimist" type narrative.

[(Anonymous)]

I need to sit next to you.

[jack]

Ooh, you did the interesting thing again.

I'm not sure if there's any interesting topics we haven't already talked about. There are some difficult ones, like "how much should you believe in God" and "how capitalist should you be", where I don't know how much we differ vs. how much we have different background and terminology for similar ideas.

I haven't formed informed opinions about smacking children. My default opinion inherited from my parents and friends who are parents is "ugh, just never". But I've never been in a situation where I would need to think about it.

My little-formed opinion is that you shouldn't hit anyone just because you're angry, and when you're angry, you probably can't readily judge if this is a good exception or not. But are there any times when it would make sense to plan for it as a necessity? And I feel like, *in theory* there might be, but I've not seen them.

Or to put it another way, it's counterproductive, unhelpful and wrong to hit dogs, and it's counterproductive, unhelpful and wrong to hit adults. Is there something about children that makes them the exception? It doesn't seem likely to me that's the case! But it's possible there is and I don't know it.

[rysmiel]

I think the fundamental underlying problem with a lot of the AI risk stuff is that it seems to take for granted that something twice as smart as us will find it easier to double its intelligence than we would, and so on exponentially, which even without getting into the relevant definition of intelligence, really makes very little sense to me. The notion that that process will either get easier as it scales, or be a smooth continuous curve, smacks of people reading Vernor Vinge's Singularity-based SF and taking this as an insight into Truth rather than a cool concept for telling stories about - and it is kind of annoying how much one actually has to specify why the Singularity is not an issue, in some kinds of SF these days, which feels kind of like having to pre-emptively justify that one's particular fictional setting as not young-Earth Creationist to me.

[woodpijn.livejournal.com]

We can't double our own intelligence unless we learn how to engineer biological brains. But a software-based AI only needs to know how to program if it wants to double its intelligence

[(Anonymous)]

Trouble is I don't have a very good sense of what you might disagree with me about, that I might want to persuade you to change your mind about...

[naath]

I used to have all sorts of ill informed opinions, I've aimed to become informed, in some cases changing my position and in other cases not (but now having the position for a better reason).

I think there are lots of things out there that I'm horribly ill informed about and becoming informed would maybe change my mind about my ill informed biases (I try hard not to be too *attached* to ill informed biases) It would be harder to convince me that an opinion that I felt was actually well founded in study of the topic was wrong I think, you'd need actual new facts or a real new perspective.

I think my most controversial opinion is probably fur. I think that if you are happy with killing animals to eat them, then killing animals to wear them is no worse (nor better), and certainly that leather is no different to fur morally speaking. I have no special sympathy for cute furry (vicious) critters than for cows. And yes, I wear it although I can't afford much; which I feel is morally consistent with eating meat and wearing leather shoes, and environmentally better than fleece which is made out of plastic (although I wear that too, since I have some).

[rysmiel]

I think that if you are happy with killing animals to eat them, then killing animals to wear them is no worse (nor better), and certainly that leather is no different to fur morally speaking. I have no special sympathy for cute furry (vicious) critters than for cows.

I feel a difference, which may not be rational, between wild animals and animals like domestic cattle which have been artificially selected to co-exist with humans ad serve human needs for long enough as to be a long way from viable in the wild; the latter seem to be at some intangible level much more ours than the former. I realised this after a few years ago seeing a rather awesome snakeskin jacket in a second-hand store (very Casanova Frankenstein) and realising I'd feel a lot more comfortable with it if I knew people had eaten the snakes afterward. (I've never eaten snake, but I've liked the couple of times I have had alligator.)

[obandsoller]

Belief change: "There is one universal theory that explains everything, and we should work towards finding that." - - > "Not even sure that we will ever unite quantum theory and general relativity, because I'm not even sure there is an objective truth to the questions being asked. And we probably need a patchwork of different theories and use what is best for each situation."

Belief change (the first thing here isn't exactly a belief): something like "My sisters are being awkward and weird about people who are just friendly" - - > "Street harassment is a thing and I should listen a bit more to people complaining about a thing than listening to people who want to dismiss what they're saying"

[jack]

I don't know if we will ever *reach* a single theory, but can you unpack more why you think QM and GR are not going to be reconciled? I really don't understand enough about it, but I'm still at the stage of just assuming they will (or could be) eventually. Many of the theories proposed seem very weird and doubtful. But surely just ordinary particles are subject to both QM forces and gravity, and must obey *some* rules? And mostly they obey one or the other because the difference in effects from the other compared to pre-20C theories are negligible. But surely there must be some rule which describes what those particles do?

I assume you have a good reason for not thinking that, but I don't grok it yet.