ineptech

Why is it reasonable to expect unbiased results from an LLM trained on internet content which almost everyone agrees is biased? Isn't this the expected outcome?

If anything, I'm surprised the results are as close as they are. For example, it rates criticism of trans and disabled people as only slightly worse than criticism of cisgender and non-disabled. If this discrepancy were (as some in this thread seem to be suggesting) the result of some liberal OpenAI employees intervening to favor their own side, I'd expect those bars to be much farther apart.

I agree, but I do worry about it being ruined some other way - forcing me to use Chrome, censoring emails, bundling it with a paid service, ad-blocker-blocker, something else...

I get the "This app is not available for any of your devices" that Play store very unhelpfully never explains. Any guesses why?

Just stop making sprint commitments. If your priorities change from day to day, what use is a list of priorities from two weeks ago?

Thank you! Once I saw this comment, very satisfying puzzle. At first I was fumbling my way through randomly, but around halfway I started to visualize the board as two "sets" of connected paths and grokked how to switch between them using the lower-right quadrant.

It's not about making me safer, it's about keeping offline OSes as an option. The ability to buy a general purpose computer and do arbitrary things with it is something we take for granted, but will never get back once we lose it.

I generally agree, but Mastodon is getting a lot of dev attention right now due to the Twitter problems, and while FOSS tends to improve slower and more incrementally, it also tends to only improve (as opposed to user-hostile pivots and so forth). For this use case, I'd rather use a clunky service that'll be around for the rest of my life than a slick one that might get bought by Elon Jr or whoever.

Fair enough about the not dredging up comments thing, I shouldn't have piped up, but you're taking this as waaay more of an attack than it was. The more nicely-phrased version might've been something like "If you're only casually acquainted with this debate, and are thinking of reeling off a low-effort paragraph or two of rebuttal, it's probably not going to go anywhere because his position on this famously contentious and vitriolic topic is well-studied." It wasn't phrased that nicely, because I like and respect Scott and find these threads a little triggering, but it was a mildly rude comment from the sidelines, not an accusation of devilry or a claim to be refuted. The "weird fixation" and "bogus argument" stuff is a little over the top.

I was referring to this[0]. I said you've felt that way for a long time, not that you post about it frequently. All I'm accusing you of is having held this position strongly and for a long time. If you say the people trying to convince you otherwise aren't wasting their time, fair enough, my mistake.

0: https://news.ycombinator.com/item?id=16560938

This is egregiously false. The author wrote a lengthy rebuttal of neoreaction[0]. The only sense in which he is sympathetic to Yarvin is that he reads his essays and argues against the ideas in them, as opposed to quoting the two worst sentences and calling him racist.

0: https://slatestarcodex.com/2013/10/20/the-anti-reactionary-f...

Are Chrome accounts separate from Google accounts generally, i.e. will the passwords go poof if your gmail account gets banned for mysterious reasons?

> I also suspect that, much like RVs or the apartments units in the Nagakin Capsule Tower, repairability and modifiability have been compromised in the design.

I think it makes more sense if you think of it as a bottom of the barrel option, say for jails or airports or public housing, rather than an ill-considered attempt to replace the washrooms in suburban houses. A Minimum Viable Bathroom, if you will, that could've been stamped out by the millions in an alternate timeline where the US decided to build council housing.

That evolutionary instinct came from a time when humans lived in communal support networks and new parents had help from extended family, and does not translate well to a world where Mom has to work 9-5 in an office a few months after birth. You might as well express shock that suburbanites don't supplement their diets by foraging for fruit and mushrooms.

Besides, it's not like parents 10K years ago had the option to let their babies cry it out in a separate room with a noise machine, and evolution selected against it because co-sleeping babies reproduced more. "Babies cry because they need to co-sleep or they will suffer some serious problem" sounds reasonable to me, but so does "Babies cry because it gets them more nutrition by keeping Mom so sleep-deprived that she delays her next pregnancy." Making just-so stories about behavioral evolution is dubious, the bar needs to be higher than just sounding plausible.

I think part of this is spot on - the loss of control of our data is intimately tied to the fact that many of the use cases we want require a server, and most people don't have one - but the solution described in step 6, non-profit like co-ops to buy and share server resources, seems dubious to me. It seems much more likely that we reach the future they're describing, if at all, through typical families adding a cloud virtual server to the list of things they spend $10/mo on. The amount of data we need to share with the world won't require that much bandwidth, even including social media (for non-celebrities anyway).

Other than that, very intriguing. I don't know if it's all workable but one has to start somewhere.

Wow, this has gotten bad fast. I did some reading on the water shortage in the Southwest the last time I saw a story about it on the FP, and came away with the idea that this is a classic coordination trap, and the only solutions are a) everyone involved spontaneously decides to conserve water in unison, and b) the Feds enact quick, decisive regulations that almost no one (the states involved, counties and cities, developers, industrial corporations, big agriculture, etc) likes and enforce them harshly. The former is a fantasy and I'm not sure the latter isn't either. It's hard to draw any conclusion other than: don't move there.

Previous article: https://www.newyorker.com/news/letter-from-the-southwest/the...

HN discussion: https://news.ycombinator.com/item?id=31987408

You've misunderstood me, no one is being excluded and there's no real name policy. Use a pseudonym, make two accounts, make ten or a hundred if you like. What you can't do is use the app to prove that you really are John Smith, if you didn't sign up as John Smith. And to the extent that one of the main features of the app is to prove your identity, the app would not be useful to someone who doesn't want to do that.

The whole point of vouching for someone is so they can prove they are who they say they are. If you don't want to do that, then don't do that. Or, if you have two identities, I suppose you'd have two accounts.

Anyway, things don't have to be for everyone. In the unlikely scenario in which this gets built, there would still be other websites too.

> You seem to be under the impression that real life access is the indicator of closeness of relationship.

No, just the hardest part to fake. You can copy my photo, hack my email, mimic my prose style, and steal my credit card, but you can't get the people in my school's PTA to say, "Yeah that's him" in real life.

I mean, nothing would stop you from making a fake identity, and friending other anonymous people, and posting spicy memes. But you can do that on mastodon today, right? This is for the cases where you want to prove your identity. If most people use it that way, then the anonymous people would be little subnets that are either separate from the main graph or connected by only a few nodes. Typical users would* have dozens of people vouching for them, who each have dozens of people vouching for them, etc, which is strong evidence that they're who they say they are.

* in some hypothetical future where this gets made and is wildly successful

If you don't want people to know that you're friends with Dave, don't friend Dave, or don't use it at all. This isn't supposed to replace pseudo-anonymous networks like Twitter, this is more like Facebook without all the bullshit. The use case is "I know Dave, and I don't care who knows that I'm friends with Dave, and I'd like to let him know I won't make it to the PTA meeting without involving an enormous advertising corporation".

You wouldn't be first-tier friends with people you only know from the internet. You'd be first-tier friends with neighbors and cousins and the other parents at your kids' school. Then when I prove (keybase style) that I, HN user ineptech, am also "ineptech" on this social network, you would know that I really am a 48-year-old guy in Portland and not some other person because I have also friended all my irl friends, and our friends-of-friends-of-friends eventually overlap.

Maybe "friend" is the wrong term. Maybe "vouch". You could still message me, but you wouldn't vouch for me, because you don't know from HN whether I am who I say I am.

You only "friend" people in real life who you trust not to act maliciously. If you unwisely trust someone who is malicious, you encounter various problems, just like in real life. But it would be very difficult for a malicious person to construct a realistic-looking subnet of sockpuppets because that would require getting lots of real people to make that unwise decision.

Not saying this is perfect, but I predict that the vast majority of users who "friend" 20+ people irl who also have friends will be fine, and borderline cases will be borderline.

I imagine that some subset of users (1/10th?) would need to aggregate and serve their own and first-tier friends' content from a cheap VPS, and the Kardashians and Elons of the world might need a CDN of some kind, but if it's limited to short text posts I imagine it would be a tractable problem.

Fleeting idea that could be combined with that: a web-of-trust social network where you can only "friend" people in real life, by tapping phones.

When you do so, the app might say, "The person you just met claims to be Joe Schmoe, do you want to vouch for them?" If you approve them, they can message you and vice versa. A friend of Joe's can see you in his friend list, and try to message you, and you can accept it if you trust Joe, but they won't be a first-tier friend until you meet them. Your tweet-like posts can be seen by anyone, or your friends only, or people within N connections of you, as you prefer.

I think it could be implemented in a distributed way, with no central server, if some proportion of the users are willing to serve their traffic from a VPS rather than just their phone. If someone cheats (uses a fork of the app that lets them "friend" people they haven't met, create fake identities, lie about their friends graph, etc), it wouldn't affect you unless you trust them. Over enough time and with enough use, this might be good enough to figure out whether someone distant from you (e.g. someone you're about to make an Ebay purchase from) is using their real identity or not, as the "main" part of the overall friend graph that a real user with a lot of friends is connected to would be structurally distinguishable from the subnets created by cheaters.

(This is not a cherished idea I've been working on for years and am prepared to defend, just a random idea I thought I'd post in case it sparks an idea for someone, so be polite in ripping it to shreds pls)

> If you don't want to discuss specific numbers, then don't state specific numbers, say that "More people do X than Y" without trying to get specific about how many people do either.

No. It was a fleeting thought in a casual conversation about video games, not a wikipedia entry. Read more charitably. Assume good faith. And maybe consider not arguing with someone unless you understand their point well enough to be confident you disagree with it.

> I've said all I care to about this topic, you can have the last word[s]

Nerf Zerg!

I'm sorry dude, I definitely meant at least 200k, not no more than 200k. Source: I am the world's leading expert in what I meant. I honestly can't see how that can be misunderstood, either. You said SC2 gets 25k views, and I said, "The reason LoL gets 100-200k is..." and then went on to talk about how much better funded LoL is. How can you read that any other way than, "The reason LoL gets so many more viewers than SC2 is..." ?

> ...by an entire order of magnitude...

I chose such a low number, ironically, to avoid a conversation like this one. When you're suggesting that a number is high (as I was), it makes sense to choose a lower bound because it makes your claim stronger. Suppose I had said that LoL events get 5M - then you could've replied, "Nuh uh, here's an event that only got 1M views!" If I had said 1M, you could've said, "Well here's one that only got 500k!" I chose the lowest number I could that would still support my point (which, again was that LoL gets a lot more viewers than SC2), to be more confident no one could argue with it.

If you respond, could you please clarify what you think my argument was? You said I made up a number to make my argument more legitimate. What argument? What could I have been arguing that would've been strengthened by suggesting LoL's viewership is lower than it really is?

When I said LoL events get 200k views, do you understand that was correct? Because, in context, I meant, "at least that much and often more," not the opposite? If you had understood my comment, you wouldn't have had to look anything up - you would've known that LoL events do indeed get more than 200k views without checking. Right?

That requires considering the context and interpreting ambiguous comments charitably, as suggested in the HN Guidelines: "Please respond to the strongest plausible interpretation of what someone says, not a weaker one that's easier to criticize. Assume good faith." When you misinterpreted my comment and triumphantly declared me wrong and accused me of bad faith, well, that's the opposite of that. Anyone can misunderstand something, it happens, but you need to be willing to say, "Either that guy's crazy or I'm missing something" and default to the latter rather than the former, like I did about the NFL thing.

Also it might be worth taking a step back and considering the topic, and whether such a combative tone is appropriate for it. And while you're at it, do you even disagree with the point of my comments, as opposed to the numbers? If so, I missed the part where you said so.

> Then why did you bring it up?

I brought up the NFL in response to you saying, "As I write this, there are [not many] people watching SC2 [on twitch]." I meant that casually checking Twitch can give a false impression of something's popularity. I wasn't suggesting the NFL was unpopular.

I also thought I made it pretty clear that my response to your comment about NFL viewership was that we were likely misunderstanding each other, as opposed to accusing you of being wrong. Please consider doing the same next time. This has been a very unpleasant conversation.

That seems unnecessarily combative, considering how minor the thing we're discussing is.

But since I'm being called out, a) I meant regular LoL events get at least 200k views; I was emphasizing how high it is compared to SC2, so pointing out that the championships got even more is not disputing that, b) I don't know how much Blizzard spends on tourney money, nor Riot or Valve; but I do follow SC2 casually, and the consensus view among the people who would know about Blizzard certainly seems to be that it has dwindled, and c) I looked, the NFL's twitch channel has 62k subscribers and hasn't streamed in months, but I didn't say "Gotcha!" because when I see something that seems way wrong I like to default to assuming that either I misunderstood or they misspoke, and in either case it hardly seems relevant to this discussion.

I admittedly did no research, I am just (accurately, I think) passing on the view of the community, which is that Blizzard has all but abandoned the game. How sure are you of those numbers, in particular how sure are you that they're not including the "value" of the logo, and/or money that actually came from sponsors? FWIW, at your urging I did do a belated search, and the first source I saw shows 1.9M as the entire prize pool for ESL 20-21, not Blizzard's portion. It can't all have come from Blizzard, unless they plaster "Intel" all over everything just to be nice.

Look, maybe I undersold it, maybe Blizzard does chip in a non-trivial amount of prize money. But would you agree with my point, which is that Riot spends way more than Blizzard on the trappings of esports, and also that Blizzard used to spend a lot more than they do today? Who do you think pays the LoL casters, and who do you think pays the SC2 casters? And would you also agree that one day LoL's popularity will wane to SC2-like levels, and that the most likely outcome is that it will more or less resemble SC2 in the sense that it is still a real esport with real tournaments and people playing it professionally, but would constitute a "failed esport" in the sense that this article is portraying esports as failing because they don't pack arenas like regular sports?