I’m not sure what exactly you want shared, or why. Claude will say different things when you ask it the same question. This is normal and expected behavior.
This is the prompt I used (in a fresh chat, using Sonnet 4.6 in Adaptive mode):
I asked Claude again in a new fresh chat, and got a different answer. This time I had copy pasted the quote from the first Claude prompt, which left quote indicator out, so the prompt was slightly different. This is the relevant section of my second answer:
And then I tried again, this time without an extra line break between the question & quote, and with > showing up as a quote indicator (I used > as a quote indicator the first time, it just got formatted differently, I think because of the extra line break), and got another answer:
I don’t think the difference was just due to the slight differences in the way I quoted though. I have also asked things identically and gotten different answers.
I also did not clarify to Claude that the thing you were responding to did include literal genocide by non-western men. And I was using Sonnet, not Opus.
I wasn’t trying to imply that, but I do want to answer you more fully so I’m going to answer the question is genocide nonwestern? too.
Since ‘western’ is an overloaded term I’ll answer twice for two meanings. I think both are reasonable within CF and that my statement was simple enough to work with both.
intended and common meaning: western just meaning UK/US/EU/CA/AU/NZ (not sure if I’ve forgotten any that ppl would commonly consider western. Israel? but that’s maybe contentious and I wasn’t thinking of them anyway).
western in the sense of western values and the enlightenment: the specific values responsible for our current golden age (as BoI puts it). I think I got this from BoI but it seems like the idea that western culture is special is more widespread albeit less philosophical.
In answering your question:
No, for the common meaning of ‘western’, claiming genocide is wholly nonwestern would contradict nazi death camps. I would not be surprised if there are other historical examples of large-scale organized state backed genoicide from countries part of ‘the west’. I’m not aware of any current examples (unless you listen to TRAs’ claims about being genocided.) FWIW I have visited Auschwitz and I think that anyone visiting the EU close enough to Krakow should make the effort. It is not pleasant but it is important and I still think about it many years later.
Yes, genocide contradicts ‘western values’ as I understand it and how I remember it being described in BoI.
My actual point was picked up by my own ‘ask claude’ experiment, so I’ll just show you both at once (via screenshots rather than sharing the chatlog b/c anon):
shared: Same thing as me: full prompt and full response. Mostly I’m curious. I think you shared that or close to but haven’t looked closely.
I have not observed LLMs saying wildly different things from the same prompt (I understand that temperature is usually set to like 0.6-0.7 on chat interfaces so not maxed out but reasonably high).
FWIW these days I think you’d get a lot more of a difference from memories, thinking, model choice, and settings like personality and writing style.
I noticed Holodomor and will say briefly: yes genocide, but I don’t count russia as western (maybe I should?)
I wouldn’t use BoI as a source for what “western values” are. I think that there are some deeply misognyistic and racist things that were part of the TCS community, and some of those came specifically from DD.
I shared the full prompt, and the part of the responses that I got the Claude quote from (and then later, the part that was analogous to that). I did not share the full Claude responses.
I think the biggest problem with your prompt was that you cut out all context. IME it’s fine to quote a specific thing from a body of text provided that you also provide the text.
Did you mean to cut out all context? Did you think that would provide a better answer?
I think there’s another thing that is important too which is otherwise unstated: the left hates the west. That’s why I included the western/nonwestern part at all, because I think it’s very possible that Doug and his wife also hate the west, though the only evidence I have for that is that he’s left aligned + didn’t make the distinction.
Your claude does bring up a good point: the fact these aren’t parallel categories. Claude seems fixated on historical stuff which makes the asymmetry important. But if there are no western countries interested in doing genocide recently, then I think the lefty interpretation is quite uncharitable towards me. I am saying that some nonwestern people think genocide is good and some nonwestern people think genocide is bad, so my framing explicitly includes the good people and separates them from the bad (good/bad within the genocide framing).
Though I need to clear up something that I was unclear about originally. The “first line” thing. I chose this because it was reducing the scope of my claim to something I was more confident about (compared to eg using it as plan E). I still have broad negative feelings towards common attitudes towards abortion.
Quick note: I’m trying to be more conscious all the time of the fact that social media provides little-to-no reliable indication of frequency or incidence rate or that kind of things. Lily’s example doesn’t say anything about how common these attitudes are or how acceptable people find them outside the basic facts like Lily and Miquita are employed, didn’t destroy their careers by saying these things, etc. Some platforms do provide some info but are a biased sample (reddit, X), but most do not allow you to do any meaningful analysis and trap you in a bubble (yt, insta, fb, tiktok, X again, etc).
I do not think it is common that women use abortion as first-line contraception. I think it is much more common that a woman will as a one-off have sex without any plan more than pull-out + fallback abortion. Obviously it takes two to tango so this is also the man’s plan in those cases.
I have other thoughts, but just wanted to quickly say - pull out with “perfect use” is estimated to be 96% effective. So that’s actually more effective than many FDA approved methods, like sponge, diaphragm, cervical cap, female condoms, spermicide, and phexxi.
I was thinking about pull out effectiveness just now too, and from extrapolating my own experiences (including inferences about women I have known) I figured it must be on the order of 99% effective. So 96% sounds reasonable. I am surprised that FDA approved methods would be lower than that, at least if they’re lower by any substantial margin. Even still, a 99% effective method used over a year with a long term partner would mean a reasonable chance of getting pregnant.
For anyone following along, it just occurred to me that this is a good opportunity for calibrating intuition of probability. This is something I try to do every once in a while when I come across something like this.
Q: Guess only based on intuition the likelihood of an event occurring at least once if it has a 1% chance of happening per trial and N = 100 trials. Answer should include uncertainty (eg. 3% chance += 1%)
My Guess
50/50 += 25%, slightly biased towards not happening. (so with 99% effective contraception and 100 uses, something like 55%-65% effectiveness across all trials)
Working
p = 1.0 - 0.01 = 0.99 \text{\;\;\;convert to avoid issues with multiple incidences}\\
p^{100} = 0.366032341
okay so I had the bias wrong. estimate of magnitude from 0.5 was about right, just.
I just realized that I often think of these things (probability intuition) in a frame relative to 0.5 + offset. There’s a bucket for problems that fall into this method then I have other buckets for other kinds of probability problems. Is this common? Do any of you do something similar?
Hmm, it seems like this would have implications for actual estimates I’d make. I won’t post that just yet to avoid spoilers if anyone else wants to try. (edit: nvm my assumptions were wrong so results don’t apply)
I’m kinda busy, but just wanted to quickly say - 99% effective birth control means that in 1 year of use, 1 out of 100 women/couples will get pregnant.
Originally I read it a bit loosely and was thinking about ~all cases with some exclusions, but on rereading I should have read it specifically as a man pushing for an abortion when the women doesn’t want to and they didn’t use protection.
Also, I have not been counting pulling out as a valid first-line contraception. So pull-out + abortion (in my head prior to this) was abortion as first-line. But that doesn’t really make sense; almost everyone knows about pulling out.
Oh! okay, I thought it was per use. And thank you for pointing this out.
I’m surprised that pulling out is anywhere near 96%, but in that case “perfect use” seems kinda arbitrary. Or maybe it means something like at the first relevant sensation the guy pulls out. IDK how you measure that scientifically though. I think it’s highly likely that the vast majority of pulling out is imperfect, though, so it’d be useful to have data around actual effectiveness based on reports of primary contraceptive method (I haven’t looked).
I wonder if this has an effect on how I interpret data? Do any of you automatically break down graphs and shapes like this? Do artists with no math background do the same kind of thing? Is it common for this intuition to be taught in high school alongside calculus? If everyone could do this and did it, would any public debates change in any substantial way?
It seems odd to me that you linked a one minute video and then also misrepresented/misremembered its contents. They don’t agree about that. Lily Allen says she used to think it was romantic, when she was younger and presumably more confused or naive. The other woman says it is romantic, and Lily says she does not think it is romantic. She does not explicitly say exactly what she thinks it is instead, but implies it is something like pragmatic/self-serving/practical because kids will cost the guy way more than the cost of an abortion.
Also, when you quoted it as her “not remembering” how many she has had, that brought to my mind that perhaps she has them like, multiple times per year. But her guess in the video is she has had maybe five abortions in total, so that is a lot less than one per year of active sexual activity. According to Wikipedia Lily Allen is 40 years old. If she started having sex at 20 (which makes the math nice and simple, IDK when she actually did), then she averages 1 abortion every 4 years. Do you think that sounds about right for someone who uses abortion as “first-line contraception”?
One more thing: check out the comments on that video. Which do you think is more problematic? Lily Allen having five abortions, or the top comments?
Do you consider the US and the UK to be part of “the west” throughout their history, or only recently? What about other western European countries such as Spain or Portugal or Holland? If so, what about the various genocides that are claimed to have been committed by them?
For example, the treatment of various indigenous peoples in North and South America. Are you unaware of those examples? They seem a lot clearer to me than something more controversial and topical like mistreatment of trans people.
Edit: I re-read and I think your mention of trans people was in context of “current” genocides, so that makes more sense to mention. So does that mean you are aware of and would agree that historically there are lots of examples of genocides perpetrated by “western” countries?
As the video mentions, women often spend money on the household and kids. A lot of shopping (food, clothes, soap, school supplies) is productive work that someone needs to do.
Do you have specific types of purchases you’re concerned about? You could look up stats or articles about them and then share your conclusions.
When talking about birth control, the “perfect use” effectiveness basically means how effective the birth control method is if everything goes right, and there are no user errors or other factors that reduce the effectiveness of the method. It’s worth noting that most places have actually stopped giving “perfect use” numbers for birth control. E.g. both the FDA and Planned Parenthood list the typical use numbers of birth control methods.
“Perfect” pull out would mean that withdrawal happens far enough ahead that ejaculation happens completely outside of the vagina and away from the vulva. (Also if there was a previous ejaculation, the man must urinate between them.) So if a pregnancy happens, that means it happened from the pre-ejaculate. The 4% chance of pregnancy per year is basically saying that you have a 4% chance of getting pregnant from pre-ejaculate.
For couples who use this as a serious birth control method, usually the man withdraws far enough ahead that they have to finish manually. Also, there are many couples who have tested the man’s pre ejaculate by looking at it under a microscope to see if there are any sperm in it, since there is some evidence that a man’s pre-ejaculate is consistent in whether or not it contains sperm. (But even if it does contain sperm, the levels are low enough that the chance of pregnancy is going to be much lower than without pulling out.)
Just want to clarify that I’m not trying to say that someone should rely on withdrawal if they know there is sperm present in pre-ejaculate. I mean more that it would explain why some people are able to use withdrawal for years successfully, and then still end up getting pregnant from “perfect” withdrawal. I don’t actually know what the numbers are for chance of pregnancy when there is known to be sperm present vs not. I just know that the overall population estimate is that 4 out of 100 couples will get pregnant with a year of “perfect” withdrawal use.
meta: going to break down the process of figuring it out. I was unsure of my intuited answer. Extra detail is included.
Intuition: “people use abortion as first-line contraception” (or “abortion as first-line contraception” depending on context.)
I think the “that” you’re referring to is this one (emphasis mine):
It’s relative to “when they do [that (2)]”, and is the same “that”. An alternative but equivalent sentence would be Also, when they do that, how often does the man push for it while […].
“That (2)” refers to the prior question:
From the question, that refers to something people do (inferred from context) and the only thing that fits from the prior question is “abortion as first-line contraception”. So that’s what I’d choose.
Unwinding the stack a bit: I think it’s a rare problem that people literally use abortion as the first-line contraception.
