MathWizard
Good things are good
No bio...
User ID: 164
MathWizard
Good things are good
21hr ago
Has anyone done work for Data Annotation or other similar online AI labeling jobs? I have a PhD in math, and have spent the past few years doing mathematical modeling in Postdocs only to realize that I don't really like writing and publishing papers. Some combination of not feeling like the work matters, getting bored of working on the same project for a long time without any feedback, and then eventually finding out that nobody thought my paper was interesting. Mehhhhhhh. And then I lose motivation and do lower quality work and my next paper is worse. I need to get out of academia. But I also don't really know what else I want to do. I'm good at math. I'm decent at programming, but I don't have experience making truly functional consumer-facing apps, all of my coding has been mathematical models that I run myself and keep tinkering with to add features whenever I want to experiment with what happens when different features or parameters of the model get tweaked.
I'm also settled down in a medium-sized town with existing but limited local career options. I have a house, and a wife who is very attached to her job and family, so remote work is vastly preferable. I'm also pathologically terrified of getting stuck in a boring 9-5 office job that eats my life away. I very much like the flexibility of working from home.
So... at least for now, Data Annotation looks promising? The advertisement claims that it pays $40/hr for Math and Programming talents, which I think I can do (unless they're super ultra competitive and only give the good work to people better than me?). The internet consensus seems to be that it's not a scam, but you might have trouble getting enough work to do it full time. And I could work my own hours, and work on discrete completable projects that feel more gamey and give feedback.
Does anyone have direct experience with this and can provide a more accurate and detailed account? Also, I think there are a couple of other similar companies that do this, so I'm not sure whether I should apply to one of those instead if they're better somehow. Or if I should apply to multiple and split my time between them in order to get a better pickings of the higher paying work? Or do you just anti-recommend the entire thing because it's not worth it? I'd like to hear thoughts and opinions from people who have either done this or know people who have done this, or know of similar remote work for someone with my talents.
MathWizard
Good things are good
1d ago
No, it's the limit on the amount that Carlos's insurance would cover.
My bad, that's what I meant to ask but got their names mixed up.
To clarify: This situation arose solely because Keith was impatient. He asked the two judges to impose the 200-k$ limit because he wanted to get his money ASAP, without waiting for the bankruptcy proceedings to finish.
But why was that necessary? Shouldn't there have been an option to do the thing he tried to do? That is, have his case proceed but, because the accident occurred prior to the bankruptcy, anything that exceed the insurance value can retroactively be voided by the bankruptcy? Or is that not possible because it would make him a creditor and the bankruptcy has to figure out how to pay those out?
Wouldn't it make more sense to put a hold on the bankruptcy proceedings and handle the lawsuit first?
MathWizard
Good things are good
1d ago
If it's well-designed then a good run that gets cut short only due to scaling should yield a huge amount of meta-currency and reward you with faster progression. There's nothing that kills a roguelite for me faster than winning on literally the first try because of some combination of luck and the game being too easy on the base difficulty.
I think the main problem is that Roguelites are appealing to two different demographics simultaneously. There are the hardcore gamers who want to challenge their wits and skills and slam their heads into a wall over and over again until they get it: people who play Souls games and lots of multiplayer PvP and brutally unforgiving games, and Roguelites are often good at that. And then there are more RPG-leaning gamers like me who want to grind out levels and currency and overcome challenges through a combination of skill and tenacity, with the ability to fungibly trade one for the other. Skill should be rewarded, but skill and progress should both grow concurrently until the sum combination is enough, so that I can take risks without failures being a literally pointless waste of time with nothing to show for it. And also have an endlessly increasing difficulty so that through progress and rewards I can eventually tackle and overcome higher and higher challenges that used to be literally impossible from the beginning of the game. If the hardest challenge of your game can be beaten in 1 hour by a player of sufficient skill level, then once you reach that skill level the game has no replay value. But if you never reach that skill level then you can never clear the game no matter how hard you try. In my opinion. I understand that lots of people have different preferences than me. But this is the weird sort of interplay, where roguelites are (trying or accidentally? not sure) appealing to both types of players at the same time under the same label. So a lot of roguelites throw some token but short and unimportant meta-progression in there and just scale it so the hardcore players can quickly unlock everything and then balance the game under that assumption, which partially satisfies but partially annoys both types of players.
MathWizard
Good things are good
2d ago
Is $200k the limit on the amount that Keith's insurance would cover? If not then Keith is entirely in the right here, and the bankruptcy judge should not have required the limit of 200k in the first place. There's no reason the amount should have been relevant for this ruling except if it would be dis-chargeable via the bankruptcy.
If it is the insurance limit then the best outcome would be the $1.6 M being the official amount awarded and the $1.4 M being retroactively discharged by the bankruptcy.
It's unfortunate that the laws aren't smart enough to do the obvious thing.
MathWizard
Good things are good
2d ago
I generally think time limits are bad in games, with a few exceptions. The fundamental problem is that it's a threat. It's a threat that, if you do poorly, you're going to have to restart the entire game from scratch. Like a final boss that, if it kills you, deletes your save file (though less volatile). I don't want to get 90% of the way through a 20 hour game only to have to start over from scratch. I rarely play games a second time unless they are exceptionally good, I'm not replaying the entirety of your game over again just because I wasn't quite good enough the first time.
The main exception I have to this is if there's meta-progression, like in Roguelites, or like Dead Rising. If you've got a 1-2 hour turn around, and I unlock new stuff every time, and the entire game is built around randomized content so it's not just the same thing again, then we're good. Or like in Dead Rising if I get stronger and it's basically a new game plus where I can solve all the problems that happened the first time around there's wayyy less risk of failing the second time around, I can take that. What I don't want is the game to tell me that the last 20 hours of play time were pointless and none of it counts for anything.
That said, I didn't have to replay Pikmin 1, because I didn't fail. If the time limit is generous enough then the majority of players don't run afoul of it. The threat looms in the background, but isn't implemented. If it's set just right then it creates stakes and pressure: the player has to act strategically and not mess up and get their party slaughtered too many times or it'll take too long to repopulate, so it feels more important to perform well. But if it's too generous then players don't feel this pressure and the time limit might as well not even exist. But Pikmin 2 was able to have a lot more content in part because of the lack of a time limit: you can keep playing the game after you "beat" it and go explore and get every last piece of treasure because there's nothing stopping you from continuing to play.
I have not yet played Pikmin 3 or 4, so I can't comment on it there, though I intend to eventually. I anticipate that the time limit in 3 will either be obnoxious if its strict, or superfluous if it's easy. There's very rarely middle ground.
MathWizard
Good things are good
5d ago
That never held water. All people, regardless of their sex or sexual orientation, can marry someone of the opposite sex of any sexual orientation. Gay men are just as free and equally allowed to marry a woman as any straight man. If the gay man doesn't want to marry a woman, that's his choice, but he's legally allowed to.
And pretty much all of the equality under the law anti-discrimination stuff has carveouts for compelling state interests. Like, say, bearing and raising children and ensuring the survival of the species.
Telling gay people that it's illegal to have sex with each other would be one thing: the state intervening in a place where it has little compelling interest or jurisdiction (an argument could be made about preventing the spread of STDs, but it's weak, and promiscuous straight people do that too). But marriage, at least from a legal perspective, is a privilege the state recognizes for people to incentivize the formation of healthy and stable families, which gay people do not do. Arguing it's "equal protection under the law" is like arguing that childless people should get the same tax deductions and/or welfare aid as people with seven children because otherwise you're discriminating against the childless.
MathWizard
Good things are good
6d ago
Treating people kindly and with love and trust is always the solution to any is-ought problem in any culture I've been to because it absolves yourself of the guilt of having acted unkindly or unlovingly and if someone interprets it incorrectly it is not because your underlying intentions were wrong.
This is only true if you tautologically define the term "kindly and with love and trust" to contain all of the complexities and nuances of the broader "is-ought".
Is it kind, loving, and trustful to lock your house or your car? Well, it's kind to the people inside the house, less kind to the thieves that want your stuff.
Is it kind, loving, and trustful to guard your wallet from pickpockets in a crime-ridden area and stop one if you catch them mid-theft?
Is it kind, loving, and trustful to punish someone for a crime? You can argue that it's kind to the victim, but unkind to the perpetrator being punished. Or you can make a complicated argument about how it's ultimately "kind and loving" to the perpetrator because the punishment will help them learn the error of their ways and become a better person which will ultimately be for their own good.
I'm not saying generally acting with kindness, love, and trust is wrong. They're good guidelines when to look to when trying to ground your decisions, but those words alone do not automatically solve all of the potential ethical dilemmas and tradeoffs inherent to the complexity of the real world.
MathWizard
Good things are good
8d ago
How easy is it to smuggle a nuke, and how long would it remain viable once smuggled?
My impression of nuclear prevention and watchfulness is that it takes a lot of science and infrastructure to refine uranium into a usable state for weapons, and you can't really build all of that stuff without a lot of commotion that foreign powers will notice.
But lots of nations already have all of that infrastructure, and the country wanting nukes only needs the end product. If country A without nukes allied with country B with nukes, would country B be able to use their own infrastructure to do most of the work and then secretly pass them enough weapons grade uranium and/or assembled nuclear warheads to stick into missiles that everyone else thinks are non-nuclear? And then several years later when it became relevant they announce "Tada! We have nukes!"
Or would this be immediately caught while happening and result in massive international penalties for countries A and or B?
MathWizard
Good things are good
8d ago
The ideal temperature for human comfort is around 20C, which is why people set their thermostats around there. Anyone setting their thermostats to anything meaningfully distant from ~20C is doing it to save money. If you're outdoors, you maybe want a bit more if it's windy, or a bit less if it's sunny or you're doing a lot of physical activity, but you want the sum of all effects to average you back so your individual subjective feeling is around 20C.
Whatever combination of sunny/cloudy/rainy gets you closer to 20C is the ideal weather for your region.
MathWizard
Good things are good
10d ago
On an economic level, I agree. This is a very very hard problem. In the case of orphans, you can resolve it by sending them to foster families or other forms of government housing instead of just handing a check to reward every kid that runs away from home. The state still pays lots of money, but the kids don't directly benefit because they get substitute parents instead of just money. But you can't really do that with single parents. You can't realistically assist single mothers with state-funded foster-fathers who come and act as the missing parent for the kid. Because she's an adult and has rights, there's a lot less coercion and control that you can't use to force compliance in the same way you can with a runaway teen (and if you tried it would turn out horribly dystopian). So we're kind of stuck handing out checks and trying to make them exactly the right size: not too small or the kids suffer poverty and neglect, not too large or the mothers have more kids and avoid marriage.
On a social level, there is so much more we could be doing to incentivize marriage. Stay at home mothers used to get respect and praise for their parenting. Single mothers used to be shamed and looked down on. Now we do the opposite. People respond to economic incentives, but they also respond to social ones too. Even if money incentivizes more single mothers, turning the dial on the social pressures in the opposite direction could help mitigate this.
MathWizard
Good things are good
11d ago
Do you believe that it's actually truly subjective? As in, it's okay for someone to kill someone else as long as they don't consider the victim to be a person? There's absolutely nothing wrong with people slaughtering "non-persons" as long as the non-person is sincerely believed by the slaughterers, and if people go around doing that you will have no complaints?
Or do you perhaps have a more nuanced and less genocidal belief about personhood grounded by something beyond mere subjectivity?
MathWizard
Good things are good
11d ago
Framing things in terms of "pro-single-mother" vs "anti-single-mother" makes about as much sense as being "pro-orphan" or "anti-orphan". You can believe that a situation is bad to be in and therefore want to help people who happen to be in that situation AND try to prevent people from falling into that situation AND not Goodhart the numbers by killing them.
DO: Help kids with no parents with money and support structures (without actively incentivizing the status)
DO: Try to prevent people from becoming orphans.
DON'T: Reduce the number of orphans by killing them
Really, a child of a single parent is just a half-orphan. Therefore
DO: Help single parent families with money and support structures (without actively incentivizing the status)
DO: Try to prevent people from becoming single-parents.
DON'T: Reduce the number of single-parents by killing them (or the children)
All of this follows trivially from the quality of life the child can expect, on average, in each state:
Full family > Single Parent Family > Orphan > Death
Whether you want more or fewer single parent families then depends on which direction you're coming from. Trying to pin people down into "pro" or "anti" single parents only makes sense if these were terminal ends rather than proxies for quality of life.
MathWizard
Good things are good
11d ago
A good rule of thumb to predict a pro-life person's opinion on something is to mentally replace the fetus with a 1 week old (post birth) baby. Or, if you don't think babies should have rights either, maybe a two or three year old. That is the logical implication of believing fetuses are people.
Would you have sympathy for a mother who killed her 1 week old baby because her husband committed suicide? Would that sympathy extend far enough to excuse the act?
MathWizard
Good things are good
16d ago
I see where you're coming from. On a gut level I immediately want to retort that punishment and forgiveness should be equally affected by your smallness and anonymity. If you're just one of many people such that your forgiveness barely matters, then your punishment barely matters too, especially since the external outcome of your forgiveness would be the cessation of your punishment/shaming.
But since it's also the case that
A: Punishments are applied in a decentralized way, with each person using their own individual criteria for what should receive shame
B: The impact of punishment via shame is nonlinear. Getting 20 death threats doesn't actually feel twice as bad as getting 10 death threats, so reducing the number of shamers by 50% doesn't actually help all that much.
Your point probably stands. Aella could repent and change her ways, and maybe 50% of people would forgive her and the shame would go down, but the other 50% would continue And also probably a bunch of sex-positive people would start shaming her and it might even end up worse. So then she's paid the massive social and lifestyle costs of repenting without actually solving the shame. Without a near-universally recognized authority who can forgive her and enforce other people's forgiveness (in deed, even if not in belief), she has no incentive to repent (beyond a genuine realization of being wrong and a self-sacrificing desire to do the right thing despite the costs).
Which in turn massively decreases the pro-social utility of shame. The point of punishments is to disincentivize the punished behavior, both on the part of the person being punished, and other people who witness them. But we've essentially lost half of that. If we make her miserable enough maybe we'll scare others away from following her example, but sometimes young people are stupid and do stuff before they realize the consequences. And anyone who does and then changes their mind is just stuck in a world where they can't be forgiven. Or more likely, doubles down on the side they're already on because they know they can't go back.
I don't know that we can do anything about that. But it still kind of sucks.
MathWizard
Good things are good
18d ago
I broadly agree with you, with the caveat that I think there's a little more room for charity/empathy/forgiveness for people who are being harmful by accident rather than on purpose.
A murderer knows that what they are doing is wrong. They know that they are going to inflict a huge harm on others (both the direct victim and everyone who knows them), and they do it anyway. They are evil and they know this and they do it anyway because they are angry/selfish and care more about themselves than others. It's an issue of bad morality.
Aella believes that she is doing good. She has a distorted view of society/psychology in a way that makes her think that her lifestyle will make the average person more happy instead of less happy. She does what she does partly in order to make herself more popular and justify her lifestyle, but in part because she genuinely thinks it will make the world better. Now, likely some of this is cognitive dissonance: maybe somewhere deep down inside she knows its wrong but doesn't care, but I think the majority of the issue is a question of bad objectivity.
Now, from a consequentialist perspective Aella is probably worse and so, if necessary and in isolation, we would be willing to inflict a higher cost in order to stop her. But the norms of "have harsher punishments for people who hurt others on purpose than those who do it by accident" is useful in general, as is "have harsher punishments for people who hurt others in unambiguous ways than those who hurt others in ambiguous and indirect ways". This makes it easier for people to know what to expect and adjust their behavior ahead of time (decreasing the rates of bad behavior) instead of doing it anyway and then getting punished randomly and unexpectedly afterwards. It also decreases the ability of people to apply punishments to good behaviors by making convoluted arguments about indirect harms. It also gives more opportunities for forgiveness and redemption. There's a non-negligible chance that at some point Aella will observe more of the effects of her actions, realize her mistake, and then change and start genuinely helping people and undo the damage she's caused. Because her underlying motivation: wanting to help people, doesn't need to change, she just needs to reconcile it with the desire to be popular and slutty and realize they're opposed instead of synergistic. I suppose murderers can also change and become better people too, but it's a different kind of change and it's impossible to undo the harm they've already caused.
This is entirely consistent with the principle of "shame Aella and jail murderers" rather than the other way around. But the shame on Aella should be tempered by education and a hope that she herself learns from her mistakes instead of just having the shamers by maximally cruel to make her suffer and hate them and reject what they say.
MathWizard
Good things are good
1mo ago
(which is were they roll the credits and probably marks around 30% of the games content)
For real? Or are you exaggerating here? I stopped playing after the first ending on the assumption that I had seen most of what it had to offer and the remaining puzzle threads that I had discovered but not yet solved seemed pedantic and annoyingly completionist and I didn't think I would have the patience to grind through them just to get a "true" ending.
But if there's literally more than half the game remaining (in terms of actual content, not merely playtime spent grinding runs hoping to get lucky) then I might pick it back up.
MathWizard
Good things are good
1mo ago
The courts should do their jobs and not do someone else's job. There's no contradiction here. It's not about the total magnitude of their power, as if there's some number that should be summed up over all the things they do and try to make sure the sums line up, it's about jurisdiction. The role of the judiciary is to interpret the law as written and intended, and apply it to individual cases, which are frequently weird and contain many facts and details that might make them edge cases or involve multiple laws that need to be combined together.
If the law doesn't say a thing and an activist judge pretends that it does by inventing new definitions for words that clearly were not what those words meant when the law was written, then they are legislating their own new laws, not actually judging. If judges go to some agency run by unelected non-judges and asks them to interpret the law for them, then those people are the judges, and the elected judges are not actually judging. They're supposed to judge, not legislate, not outsource.
MathWizard
Good things are good
1mo ago
I think this is typically handwaved away by assuming that if we, as humans, manage to solve the original alignment problem, then an AI with 100x human intelligence will be smart enough to solve the meta-alignment problem for us. You just need to be really really really sure that the 100x AI is actually aligned and genuinely wants to solve the problem rather than use it as a tool to bypass your restrictions and enact its secret agenda.
MathWizard
Good things are good
1mo ago
I am not a legal expert. I don't know the actual legal technicalities of "reasonable articulable suspicion". But at least taking a common sense definition of the words, "There is a serial killer drowning babies in wicker baskets, this man is on the marina with a wicker basket" is trivially "articulable", and seems eminently reasonable to me given the circumstances. Who uses wicker baskets? If the officer had no reason for suspicion then why did he call you in the first place? My understanding was that this was to prevent officers from searching people because "I dunno, he seemed kind of suspicious", not "I have a clear reason to suspect this guy in particular of a specific crime (and being the perpetrator of the same crime multiple times in the past) for a specific reason".
MathWizard
Good things are good
2mo ago
In most non-rpgs clothing doesn't imbue stats, so you can just wear whatever you want and there's no dilemma to fix.
MathWizard
Good things are good
2mo ago
I don't believe Minecraft dungeons does this (without mods?) but I have seen games that let you have two layers of armor: a mechanical one that's actually officially equipped, and a fashion one that determines your appearance. So you can have the best of both (and I think to incentivize players to buy fancy cosmetic armor with real money). I don't remember what game(s) this was, I think I've seen it more than once but not very often.
MathWizard
Good things are good
2mo ago
I think the point is that NATO, knowing they have nukes and are willing to use them, would choose not to invade in the first place. They still die in a hypothetical world where NATO wants the world to end. They live in the world where NATO doesn't want the world to end and chooses not to invade them, because they have nukes.
MathWizard
Good things are good
2mo ago
The issue, as they point out, is that outcomes are heterogeneous. If the outcome is a combination of your decision and random noise and circumstance outside of your control, then outcome will be weakly correlated with the actual value you provide. Half of punishments and rewards will be deserved, and half will be simply responding to the whim of fate.
If your punishment/reward mechanism is long-term enough, like say the profits of a company that can accumulate over time and wash out the negatives with positives, then risky but positive expectation behaviors will work. If your mechanism is "fire any CEO who has a year with negative profit, no matter why it turned out negative" then you're likewise going to incentivize conservative behavior that guarantees the bare minimum at the cost of unlucky but smart people who take risks with positive expected value.
MathWizard
Good things are good
3mo ago
I have been using Suno AI to make music about silly events and inside jokes that amuse me. I have a song about my favorite character in Gloomhaven, a song about how much I hate snow, a song about my wife being a loot goblin in a game, a song making fun of some redneck who harassed my brother, a song about a tiny plant my wife got in a toilet-shaped pot that we put on top of our toilet, etc...
Most of them sound like real songs you might hear on the radio. Nothing super profound, but not terrible. Well, actually probably 80% of the time it's terrible, but as usual with AI you discard the garbage and retry and reprompt until you get something good. Occasionally I write the lyrics myself from scratch, but most of the time I prompt Chat GPT on a topic and then tweak the lyrics to fit the context better before giving them to Suno to make a song. It's wonderful, and I am gradually accumulating a playlist of actually good songs that mean something to me.
How do I accurately evaluate my worth? I'm too heavily confounded by impostor syndrome that I can't tell where it ends and my true value lies. I'm definitely below average for a Math PhD in terms of accomplishments. None of my grad-school work ended up getting published, and I've published 1-2 papers per year in my postdocs which have gotten ~5 citations each. I seem to work a lot less than my peers, and my advisor/bosses have been too busy and/or easy-going to push me, so I've kind of been coasting. That said, I am smart enough to learn stuff when I do try, and got a Math PhD, and know how to hack code together into something that compiles. I don't know what that's worth. What I do know is I'm not willing to put in the 60+ hour per week that the professors I've worked under seem to do writing grants and managing grad students and whatnot. At least not consistently, I would put in a couple long weeks if I really had to.
And I don't want to move, which drastically reduces my options. But on the other hand, the cost of living is not very high, and I'm currently DINK, so technically could survive on just my wife's job, but that wouldn't really be fair to her. On the work-life balance front I heavily lean towards the life part. Work is there so I don't starve and can afford people to do stuff like house repairs that I don't want to do.
I'm sure someone with my intelligence plus work ethic and ambition that I don't have could easily be making loads of money. However, given my constraints, is $40 an hour still an insult? My ideal position is remote, high pay per hour, few total hours, and meaningful/satisfying/moral, (I'm not phoning it in on a job that my employer expects more from), but I'm not sure what that is or if that's too many variables maximized simultaneously and I might need to compromise on some.
More options
Context Copy link