Posts

[LINK] Article in the Guardian about CSER, mentions MIRI and paperclip AI 2014-08-30T14:04:05.959Z

Comments

Comment by Sarokrae on [LINK] Article in the Guardian about CSER, mentions MIRI and paperclip AI · 2014-08-30T16:05:48.021Z · LW · GW

I wouldn't worry too much about the comments. Even Guardian readers don't hold the online commentariat of the Guardian in very high esteem, and it's reader opinion, not commenter opinion, that matters the most.

It seems like the most highly upvoted comments are pretty sane anyway!

Comment by Sarokrae on [LINK] Article in the Guardian about CSER, mentions MIRI and paperclip AI · 2014-08-30T15:51:34.846Z · LW · GW

I've read a fair number of x-risk related news pieces, and this was by far the most positive and non-sensationalist coverage that I've seen by someone who was neither a scientist nor involved with x-risk organisations.

The previous two articles I'd seen on the topic were about 30% Terminator references. This article, while not necessarily a 100% accurate account, at least takes the topic seriously.

Comment by Sarokrae on White Lies · 2014-02-11T00:31:57.797Z · LW · GW

This is a summary reasonably close to my opinion.

In particular, outright denouncement of ordinary social norms of the sort used by (and wired into) most flesh people, and endorsement of an alternative system involving much more mental exhaustion for the likes of people like me, feels so much like defecting that I would avoid interacting with any person signalling such opinions.

Comment by Sarokrae on Community bias in threat evaluation · 2014-01-17T12:41:26.922Z · LW · GW

Actually I don't think you're right. I don't think there's much consensus on the issue within the community, so there's not much of a conclusion to draw:

Last year's survey answer to "which disaster do you think is most likely to wipe out greater than 90% of humanity before the year 2100?" was as follows:

Pandemic (bioengineered): 272, 23% Environmental collapse: 171, 14.5% Unfriendly AI: 160, 13.5% Nuclear war: 155, 13.1% Economic/Political collapse: 137, 11.6% Pandemic (natural): 99, 8.4% Nanotech: 49, 4.1% Asteroid: 43, 3.6%

Comment by Sarokrae on Instinctive Frequentists, the Outside View, and de-Biasing · 2013-09-21T10:29:15.307Z · LW · GW

I'm pretty sure this is one of the main areas Prof David Spiegelhalter is trying to cover with experiments like this one. He advises the British government on presenting medical statistics, and his work is worth a read if you want to know about how to phrase statistical questions so people get them more right.

Comment by Sarokrae on To what degree do you model people as agents? · 2013-08-30T15:49:18.339Z · LW · GW

This post reminded me of a conversation I was having the other day, where I noted that I commit the planning fallacy far less than average because I rarely even model myself as an agent.

Comment by Sarokrae on "Stupid" questions thread · 2013-08-18T17:52:11.601Z · LW · GW

A non-exhaustive list of them in very approximate descending order of average loudness:

  • Offspring (optimising for existence, health and status thereof. This is my most motivating goal right now and most of my actions are towards optimising for this, in more or less direct ways.)

  • Learning interesting things

  • Sex (and related brain chemistry feelings)

  • Love (and related brain chemistry feelings)

  • Empathy and care for other humans

  • Prestige and status

  • Epistemic rationality

  • Material comfort

I notice the problem mainly as the loudness of "Offspring" varies based on hormone levels, whereas "Learning new things" doesn't. In particular when I optimise almost entirely for offspring, cryonics is a waste of time and money, but on days where "learning new things" gets up there it isn't.

Comment by Sarokrae on Biases of Intuitive and Logical Thinkers · 2013-08-18T10:12:41.212Z · LW · GW

As an "INFJ" who has learned to think in an "INTJ" way through doing a maths degree and hanging out with INTJs, I also agree that different ways of problem solving can be learned. What I tend to find is that my intuitive way of thinking gets me a less accurate, faster answer, which is in keeping with what everyone else has suggested.

However, with my intuitive thinking, there is also an unusual quirk that although my strongly intuitive responses are fairly inaccurate (correct about half the time), this is much more accurate than they have any right to be given the precision of the ones that are correct. My intuitive thinking usually applies to people and their emotions, and I frequently get very specific hypotheses about the relationships between a set of people. Learning logical thinking has allowed me to first highlight hypotheses with intuition, then slowly go through and falsify the wrong ones, which leads me to an answer that I think I couldn't possibly get with logic alone, since my intuition uses things like facial expressions and body languages and voice inflections to gather much more data than I could consciously.

Comment by Sarokrae on "Stupid" questions thread · 2013-07-21T20:21:23.259Z · LW · GW

"Whichever subagent currently talks in the "loudest" voice in my head" seems to be the only way I could describe it. However, "volume" doesn't lend to a consistent weighting because it varies, and I'm pretty sure varies depending on hormone levels amongst many things, making me easily dutch-bookable based on e.g. time of month.

Comment by Sarokrae on "Stupid" questions thread · 2013-07-18T10:20:58.947Z · LW · GW

I'm not entirely sure. What questions could I ask myself to figure this out? (I suspect figuring this out is equivalent to answering my original question)

Comment by Sarokrae on "Stupid" questions thread · 2013-07-17T07:24:04.618Z · LW · GW

My other subagents consider that such an appalling outcome that my processor agent refuses to even consider the possibility...

Though given this, it seems likely that I do have some degree of built-in weighting, I just don't realise what it is yet. That's quite reassuring.

Edit: More clarification in case my situation is different from yours: my 3 main subagents have such different aims that each of them evokes a "paper-clipper" sense of confusion in the others. Also, a likely reason why I refuse to consider it is because all of them are hard-wired into my emotions, and my emotions are one of the inputs my processing takes. This doesn't bode well for my current weighting being consistent (and Dutch-book-proof).

Comment by Sarokrae on "Stupid" questions thread · 2013-07-15T07:04:37.178Z · LW · GW

So I started reading this, but it seems a bit excessively presumptuous about what the different parts of me are like. It's really not that complicated: I just have multiple terminal values which don't come with a natural weighting, and I find balancing them against each other hard.

Comment by Sarokrae on "Stupid" questions thread · 2013-07-14T09:23:19.051Z · LW · GW

In the process of trying to pin down my terminal values, I've discovered at least 3 subagents of myself with different desires, as well as my conscious one which doesn't have its own terminal values, and just listens to theirs and calculates the relevant instrumental values. Does LW have a way for the conscious me to weight those (sometimes contradictory) desires?

What I'm currently using is "the one who yells the loudest wins", but that doesn't seem entirely satisfactory.

Comment by Sarokrae on "Stupid" questions thread · 2013-07-14T09:12:58.894Z · LW · GW

I can't upvote this point enough.

And more worryingly, with the Christians I have spoken to, those who are more consistent in their beliefs and actually update the rest of their beliefs on them (and don't just have "Christianity" as a little disconnected bubble in their beliefs) are overwhelmingly in this category, and those who believe that most Christians will go to heaven usually haven't thought very hard about the issue.

Comment by Sarokrae on "Stupid" questions thread · 2013-07-14T09:09:16.935Z · LW · GW

Surely a more obvious cost is the vast number of people who like tigers and would be sad if they all died?

Comment by Sarokrae on "Stupid" questions thread · 2013-07-14T09:03:08.077Z · LW · GW

I'd be alarmed if anyone claimed to accurately numerically update their priors. Non-parametric Bayesian statistics is HARD and not the kind of thing I can do in my head.

Comment by Sarokrae on "Stupid" questions thread · 2013-07-14T09:01:42.048Z · LW · GW

I second what gothgirl said; but in case you were looking for more concrete advice:

  1. Exchange compliments. Accept compliments graciously but modestly (e.g. "Thanks, that's kind of you").
  2. Increase your sense of humour (watching comedy, reading jokes) until it's at population average levels, if it's not there.
  3. Practise considering other people's point of view.
  4. Do those three things consciously for long enough that you start doing them automatically.

At least, that's what worked for me when I was younger. Especially 1 actually, I think it helped with 3.

Comment by Sarokrae on Using Evolution for Marriage or Sex · 2013-05-06T14:48:28.734Z · LW · GW

Thank you. This is appreciated. I know it's hard work, but from our point of view we can't take your word that you're not just making most of it up off the top of your head. (Also a lot of people like to independently assess the reliability of sources.)

Comment by Sarokrae on Using Evolution for Marriage or Sex · 2013-05-06T09:54:48.660Z · LW · GW

I approve of this post, everything in it seems pretty reasonable (my current OH did about 80% of the long-terming male list), though I do wish you could've added a list of citations; this is quite a lot of content to just pull out of the hat.

Comment by Sarokrae on LW Women- Female privilege · 2013-05-06T00:25:34.403Z · LW · GW

Data: Wikipedia claims E/I is very correlated with E, S/N is very correlated with O, F/T fairly correlated with A, J/P fairly correlated with C and somewhat correlated with O, and Neuroticism isn't measured in MBTI. So this backs up your claim that P/J doesn't measure any concrete "thing".

Clicking through the citation gives that N is not well-correlated with anything in men (a tiny bit with E/I), and somewhat correlated with the F/T in women. Also F/T has a small effect on extraversion in men, but it's S/N and J/P which has the effect on women.

Comment by Sarokrae on Health/Longevity Link List · 2013-05-05T21:32:23.276Z · LW · GW

Also on the "mainstream/obvious list":

Being obese is bad. Being overweight probably bad. Being underweight is probably also bad. Vitamin D good. Getting enough micronutrients in general good. Excessive red meat consumption probably bad. Excessive processed meat consumption bad. Laughter good.

(That's all I can think of off the top of my head that's not yet been mentioned.)

Edit: Oh! Forgot one. Sunburn bad.

Comment by Sarokrae on LW Women- Female privilege · 2013-05-05T20:18:02.724Z · LW · GW

The post does seem to imply that she finds understanding men easier.

Comment by Sarokrae on LW Women Entries- Creepiness · 2013-04-29T15:52:37.291Z · LW · GW

Good point. I checked by visualising a selection of people in my head asking this, male and female, with various characteristics. I had the same reaction to about equal numbers of men and women. Usually some something along the lines of "erm, can we add each other on facebook first?"

...Then again, I'm probably just particularly not-keen on giving people my phone number, and as such was reading the situation exclusively in terms of "which way of asking makes the certainty of me saying "no" less awkward?"

Comment by Sarokrae on LW Women Entries- Creepiness · 2013-04-28T23:57:10.846Z · LW · GW

If that's how you actually say it, I'd be a little concerned about how you were coming across. "Let's exchange our phone numbers" doesn't lend itself to a polite "no" in the same way as, say, "Do you want to exchange phone numbers?"

Comment by Sarokrae on Rocket science and big money - a cautionary tale of math gone wrong · 2013-04-24T15:39:45.491Z · LW · GW

and by an elementary reasoning known in physics as "dimensional analysis", dividing a number of issues by another number of issues cannot give us an ROI

This is just being nit-picky, but from a dimensional analysis point of view, both "dollars per dollar" and "issues per issue" are dimensionless figures, and are thus in fact the same dimension.

Comment by Sarokrae on LW Women Submissions: On Misogyny · 2013-04-20T23:06:08.428Z · LW · GW

Didn't see this reply as it wasn't directly to one of my posts, but I would like to reassure anyone reading that I can tell the difference between "skin crawling" and "scalp tingling", and no they are not the same thing at all.

Comment by Sarokrae on Reinforcement and Short-Term Rewards as Anti-Akratic · 2013-04-16T17:38:15.337Z · LW · GW

Some of my recent forays into reinforcement learning have been very helpful. I should point out that my life is made a whole lot easier by having a very co-operative OH who is willing to reward me or withhold reward as appropriate, so I've not needed to resort to building a robot!

Things that have been successful:

  • Every time I think about {thing I enjoy obsessing about}, I go and do the washing up. I used to have a massive ugh field around washing up, but this has quickly diminished (within days!) via association with the nice thoughts. We're thinking of applying this method to other things I have ugh fields around, since it was so quick and effective.
  • I've been doing a similar thing to D_Malik with regards to Anki cards. However, it was impractical for me to withhold a reward I would be having on a daily basis, so my OH is implementing "withhold {nice thing} unless I have reviewed my Anki cards for the previous 5 days". It's not as immediate as not eating, but seems to be sufficiently encouraging thus far.

But yeah, having a person help me do it means I avoid any sort of precommitment failure, and generally makes things much easier!

(Side note: Curly brackets clearly denote euphemisms, but I didn't want to be too crude.)

Comment by Sarokrae on Solved Problems Repository · 2013-04-12T23:36:18.897Z · LW · GW

Age, body fat proportion, maximal oxygen uptake...

In my experience, these tend to be taken into effect when calculating the "calories out" part of the equation. By what mechanism were you thinking that these mattered, that's not "calories out"?

Comment by Sarokrae on LW Women Submissions: On Misogyny · 2013-04-12T15:41:29.197Z · LW · GW

Well this is in the context of a long period of introspection on the theme of "When it comes to moral considerations, how much is my system 1 me?" The conclusion is "not very", which is one of the things I changed my mind about fairly recently. (If instinct wants to sleep with someone but reason doesn't, it is preferable for me to not sleep with them. This probably doesn't sound like a surprising conclusion, but I was confused for a long time.)

This observation was basically consistent with the way my ideas were developing, since I developed those ideas concurrently to developing luminosity. I'm afraid I can't tell the direction of causation between the two things, or whether there is any.

Comment by Sarokrae on Solved Problems Repository · 2013-04-12T12:25:27.628Z · LW · GW

1,000 calorie diets ... third group gained 0.24 lbs / day

I noticed I was confused. This doesn't seem consistent with the results of the Minnesota Starvation/Semistarvation Study. I went to Wikipedia.

Kekwick and Pawan, 1956 conducted a study of subjects consuming 1000-calorie diets, some 90% protein, some 90% fat, and some 90% carbohydrates. Those on the high fat diet lost the most, the high protein dieters lost somewhat less, and the high carbohydrate dieters actually gained weight on average. Kekwick and Pawan noted irregularities in their study (patients not fully complying with the parameters of the study). The validity of their conclusions has been questioned, and follow-up studies over a longer duration concluded that these temporary differences were due chiefly to changes in water balance (citation)

My prior consider it quite ludicrous that you can gain weight eating at a 50% deficit, no matter what your macros. The criticisms seem reasonable enough to explain the effect.

Note that the link in the citation claimed that when told to cut out carbs and eat as much protein and fat as they liked, "In all subjects, there was a reduction in calories ranging from 13% to 55% during the time they were consuming the low-carbohydrate diet."

Comment by Sarokrae on Post Request Thread · 2013-04-12T12:06:24.573Z · LW · GW

If anyone is interested, I was considering writing about using game theory in real life combined with luminosity. In particular, I was thinking of writing in an in-depth way about an example (what to do after infidelity in in monogamous relationships) for one-boxing in real life, because it felt like a better intuition pump for me than any other example I've heard before.

Newcomb's problem in real life has already been covered before in these posts: http://lesswrong.com/lw/4yn/realworld_newcomblike_problems/ http://lesswrong.com/lw/1zw/newcombs_problem_happened_to_me/

Comment by Sarokrae on LW Women Submissions: On Misogyny · 2013-04-12T11:53:14.908Z · LW · GW

I upvoted this comment for the info on porn and masturbation addiction, which is news to me, but makes sense with my model of the world, and seems to be something I ought to look into. Thanks.

Comment by Sarokrae on LW Women Submissions: On Misogyny · 2013-04-12T11:17:05.235Z · LW · GW

I believe that "skin crawling" is a common metaphor for the experience of being around someone who is creepy.

I'm just going to give one personal point of evidence which people can interpret however they want: a large part of my own understanding of "creepiness" comes from the fact that at least for me personally, "skin crawling" is actually just unwanted sexual arousal. (It took me quite a lot of luminosity practise to figure this out.)

Comment by Sarokrae on Existential risks open thread · 2013-04-12T10:49:10.442Z · LW · GW

Thank you!

Comment by Sarokrae on Existential risks open thread · 2013-04-07T18:26:51.567Z · LW · GW

Re: formatting. Try putting a blank line between bullets.

Tried, doesn't work. Anyone got any ideas?

Comment by Sarokrae on Existential risks open thread · 2013-04-06T12:46:13.646Z · LW · GW

Pushing for more democratic governments in states like Russia and China might also decrease the chances of nuclear war, etc.

How sure are you?

  • Acts of military aggression by the PRC since 1949: About 5.
  • Acts of military aggression by the USSR/Russia in the same period: About 5
  • Acts of military aggression by the USA in the same period: About 7

(I've tried to be upwardly biased on numbers for all three, since it's obviously hard to decide who the aggressors in a conflict are)

  • Wars that the PRC have participated in that were not part of domestic territorial disputes since 1949: 2
  • Likewise for Russia: 5
  • Likewise for the USA: 17

(for the USA and USSR figures I'm counting all of the Cold War as one conflict, and likewise all of the War on Terror)

Sources found here

Edit: What happened to my formatting? I've had this problem before but I've never been able to fix it.

Comment by Sarokrae on Decision Theory FAQ · 2013-03-13T21:07:45.636Z · LW · GW

I don't have a paperclipping architecture but this doesn't stop me from imagining paperclipping architectures.

So my understanding of David's view (and please correct me if I'm wrong, David, since I don't wish to misrepresent you!) is that he doesn't have paperclipping architecture and this does stop him from imagining paperclipping architectures.

Comment by Sarokrae on Decision Theory FAQ · 2013-03-13T14:42:19.932Z · LW · GW

...microelectrodes implanted in the reward and punishment centres, behavioural conditioning and ideological indoctrination - and perhaps the promise of 72 virgins in the afterlife for the faithful paperclipper. The result: a fanatical paperclip fetishist!

Have to point out here that the above is emphatically not what Eliezer talks about when he says "maximise paperclips". Your examples above contain in themselves the actual, more intrisics values to which paperclips would be merely instrumental: feelings in your reward and punishment centres, virgins in the afterlife, and so on. You can re-wire the electrodes, or change the promise of what happens in the afterlife, and watch as the paperclip preference fades away.

What Eliezer is talking about is a being for whom "pleasure" and "pain" are not concepts. Paperclips ARE the reward. Lack of paperclips IS the punishment. Even if pleasure and pain are concepts, they are merely instrumental to obtaining more paperclips. Pleasure would be good because it results in paperclips, not vice versa. If you reverse the electrodes so that they stimulate the pain centre when they find paperclips, and the pleasure centre when there are no paperclips, this being would start instrumentally value pain more than pleasure, because that's what results in more paperclips.

It's a concept that's much more alien to our own minds than what you are imagining, and anthropomorphising it is rather more difficult!

Indeed, you touch upon this yourself:

"But unless I'm ontologically special (which I very much doubt!) the pain-pleasure axis discloses the world's inbuilt metric of (dis)value - and it's a prerequisite of finding anything (dis)valuable at all.

Can you explain why pleasure is a more natural value than paperclips?

Comment by Sarokrae on Rationality Quotes January 2013 · 2013-03-12T02:02:33.013Z · LW · GW

I'm a bit late here, but my response seems different enough to the others posted here to warrant replying!

My brain is abysmally bad at storing trains of thought/deduction that lead to conclusions. It's very good at having exceptionally long trains of thoughts/deductions. It's quite good at storing the conclusions of my trains of thoughts, but only as cached thoughts and heuristics. It means that my brain is full of conclusions that I know I assign high probabilities to, but don't know why off the top of my head. My beliefs end up stored as a list of theorems in my head, with proofs left as an exercise to the reader. I occasionally double-check them, but it's a time-consuming process.

If I'm having a not very mentally agile day, I can't off the top of my head re-prove the results I think I know, and a different result seems tempting, I basically get confused for a while until I re-figure out how to prove the result I know I've proven before.

Basically on some days past-me seems like a sufficiently different person that I no longer completely trust her judgement.

Comment by Sarokrae on [SEQ RERUN] Fairness vs. Goodness · 2013-03-08T20:24:39.955Z · LW · GW

Good point. Thanks.

Link for anyone else re-reading.

Comment by Sarokrae on [SEQ RERUN] Fairness vs. Goodness · 2013-03-08T12:25:42.519Z · LW · GW

The original links in the article are dead. Does anyone have a mirror?

Comment by Sarokrae on Getting myself to eat vegetables · 2013-03-07T18:35:05.302Z · LW · GW

I agree with this. I've read somewhere (source needed) that it takes babies about 10-15 tastings to get to grips with a new, unfamiliar flavour, and I'd imagine the same principle applies for adults.

More anecdotally, both my father and my OH started off really strongly disliking the flavour of coriander in their teens, then grew to really like it after they've tasted it in a variety of contexts. My father also had this with yoghurt, and I myself with goats' cheese.

In fact, I'd suggest you start with more variety than just one vegetable, since if you attack on all bases simultaneously, statistically speaking you're going to start liking one of them faster!#

Edit: only source for my claim that I could find which cites a study is here. The study also suggests that babies learn to like a food faster if it comes paired with something they already like, which gives evidence to some the other suggestions mentioned.

Comment by Sarokrae on Tutoring Small Groups of Children (for money) · 2013-02-21T13:10:43.235Z · LW · GW

The Memrise community (http://www.memrise.com/) are quite big on that kind of thing. If you're learning a language, you can browse through their community database of mnemonics or "mems" for inspiration, and are encouraged to create your own. Their site is also quite good as a spaced repetition tool for languages.

Comment by Sarokrae on Tutoring Small Groups of Children (for money) · 2013-02-21T13:01:05.727Z · LW · GW

Good point, fixed.

Comment by Sarokrae on Tutoring Small Groups of Children (for money) · 2013-02-21T00:39:50.008Z · LW · GW

Sorry, I don't seem to have made myself clear. I was arguing against warning students against password guessing. I.e. don't remake the game, just play it as intended.

Comment by Sarokrae on Tutoring Small Groups of Children (for money) · 2013-02-20T23:22:02.474Z · LW · GW

I'd argue against this. I always saw through password-guessing as fake and not really understanding anything when I was young, but lacked the people skills to notice that the teachers and examiners wanted me to guess the password rather than demonstrate that I really understood (because I didn't understand why), and lost a few exam marks along the way to figuring that out!

Comment by Sarokrae on Tutoring Small Groups of Children (for money) · 2013-02-20T19:05:02.247Z · LW · GW

Materials-wise, I can't recommend the Murderous Maths books by Kjartan Poskitt enough. They're what got me into maths, and introduce topics at a basic level while leaving the top end open for further development. They've got lots of fun puzzles and activities. I think they're aimed at 10-14 year olds, so would be towards the top of your tutee's age range.

Less paper-based and more free is NRICH (http://nrich.maths.org/), which is run by a group at Cambridge (I've done some holiday work writing questions for them). It contains materials for all ages (pre-school up to sixth form! (4-or-5-ish to 8)) and is intended as simultaneous for teachers and keen students. It's got a good selection of mathematical games as well as themed monthly questions, and keen students can type up their solutions and submit it to them for a chance of being featured on the website. The games in particular will be a good way of introducing lots of mathematical concepts, since that's how they're designed.

In fact, teaching your kids basic game theory is probably a good idea. It teaches you them to think rationally and is easy to motivate (beat all your school friends!). Many rationality concepts and techniques are founded upon the basics of game theory!

Comment by Sarokrae on Desires You're Not Thinking About at the Moment · 2013-02-20T18:15:07.713Z · LW · GW

I've heard a common argument post-tabooing-"people" to be "I care about things that are close in thing-space to me. "People" is just a word I use for algorithms that run like me." (This is pretty much how I function in a loose sense, actually)

Comment by Sarokrae on LW Women: LW Online · 2013-02-18T13:54:12.804Z · LW · GW

Thanks for the other data points.

I always hear comments sort-of-out-loud though, the same way reading happens sort-of-out-loud-in-my-head. I don't think it's something I can switch off. I always hear tone and it would confuse me not to, even though I do sometimes get it wrong. In fact, I get confused if people I'm close to type without punctuation, since an absence of tone just registers as "the tone of being distant and brusque".

Comment by Sarokrae on LW Women: LW Online · 2013-02-17T14:17:01.762Z · LW · GW

Fair enough. I think the "I'm sorry" part of the phrase makes me hear it as not-at-all impersonal; I would leave it at "Oh, that really sucks" for anyone except very close friends.

Even with that replacement though, I think I struggle to hear the comment as sincere, because it's a weird juxtaposition of personal and impersonal: "I'm sorry, that sucks" is highly personal and fairly colloquial; the rest of the statement was more distant and formal. So even though it is coherent in meaning, it doesn't feel coherent in tone, which makes me struggle to hear it as sincere. (And when I do manage to get it sincere, it sounds "creepy" since that's what "doing social conventions wrong" sounds like)