TonesTones's recent activity
-
Comment on Harvard University Loses Student and Exchange Visitor Program Certification for Pro-Terrorist Conduct in ~society
-
Harvard University Loses Student and Exchange Visitor Program Certification for Pro-Terrorist Conduct
2 votes -
Comment on 1-1 political conversations to help mend the political divide in ~society
TonesTones Link ParentI agree that the folks who would do this in good faith already try more than most to get a picture of both sides of the aisle. I’m actually curious how much the private nature of the discourse...Such a beautiful idea but the folks who are going to participate in this in good faith are already doing so elsewhere.
I agree that the folks who would do this in good faith already try more than most to get a picture of both sides of the aisle.
I’m actually curious how much the private nature of the discourse will make a project like this more or less inflammatory than the Internet already is.
Private conversations in-person are always less inflammatory, but (in my experience) in-person interaction has this way of activating “I need this person to like me” signals in my brain in a way that virtual discourse doesn’t.
-
Comment on Sweden's recycling centres overflowing with clothes after EU-wide ban on throwing away textiles – municipalities eager to have fast fashion giants take responsibility in ~enviro
TonesTones LinkThere’s a lot of discussion in the article about how the fast fashion companies should take responsibility. Maybe this is an opportunity for me to bring up an idea I’ve had for some time. A waste...There’s a lot of discussion in the article about how the fast fashion companies should take responsibility. Maybe this is an opportunity for me to bring up an idea I’ve had for some time.
A waste tax paid by companies based on if their products go into landfill and corresponding writeoffs if they recycle things from the landfill.
The idea is that if a company sells a shirt, it’s assumed that shirt will go into landfill and the company is taxed accordingly. If the company can demonstrate that they also remove a shirt from landfill, they can writeoff and not get charged those taxes.
We let the incentive be a tradeable writeoff instead of an monetary reward so that recycling companies can form without a government middleman. Companies can pay a recycling company for the excess tax writeoffs that the recycling company is producing. Then, the recycling company is paid both to use the raw materials and paid for the actual products they sell, so that companies producing from recycled materials have an easier time making their unit economics profitable.
Looking for critiques of the idea.
-
Comment on LinkedIn executive says that the bottom rung of the career ladder is breaking in ~life
TonesTones LinkA few points I’d like to make. (A) People are definitely using AI in their jobs. ChatGPT’s monthly average users do not come from nowhere. It’s possible the 20/80 heuristic still holds and the...A few points I’d like to make.
(A) People are definitely using AI in their jobs. ChatGPT’s monthly average users do not come from nowhere. It’s possible the 20/80 heuristic still holds and the people using AI are the people who weren’t that productive pre-AI. I don’t use AI very much. It’s hard to know.
(B) There’s a lot of money to be made automating white-collar work. If it’s economically feasible (read: if equivalent compute is cheaper than a human salary), it will happen.
(C) Some tech companies are outsourcing their intro-level jobs to contracting firms, some of which are located overseas, and all of which pay substantially less. Some of those jobs exist in the U.S. and they are hiring some very bright engineers at pretty low wages. Capital will do what capital does.
-
Comment on The 'deprofessionalization of video games' was on full display at PAX East in ~games
TonesTones Link ParentI think so too. Many of the commenters here are bringing up their favorite games with 1-3 people developing. There are hundreds, possibly thousands of 1-3 person teams trying to do indie game...I guess that explains the sentiment here from what I assume is consumers.
I think so too. Many of the commenters here are bringing up their favorite games with 1-3 people developing. There are hundreds, possibly thousands of 1-3 person teams trying to do indie game development. Gamers get to play the best of them, while many teams lose their investments.
You can't specialize when you're jumping jobs every 2-3 years because people keep laying you off anyway. Leaving me at a crossroads of sorts.
I’m sorry to hear that it’s been difficult. I don’t have a lot to say to alleviate those concerns. I do want to acknowledge that it’s a hard industry, no matter who you are.
-
Comment on The 'deprofessionalization of video games' was on full display at PAX East in ~games
TonesTones LinkGame development has become substantially more accessible in the last decade. Therefore, it’s much easier for a small team or solo developer to build a game that competes with AAA titles, at least...Game development has become substantially more accessible in the last decade. Therefore, it’s much easier for a small team or solo developer to build a game that competes with AAA titles, at least for attention.
More accessibility means more competition. Lots of commenters in this thread are saying things like “games are way more fun now” and whatnot.
Competition has always been, on average, bad for owners, a mixed bag for laborers, and good for consumers.
Yes, games are better now, but largely because you can pick from more games. I’m sure there are plenty of indie game developers pouring their soul and their savings into their dream only to have it get a hundred purchases on Steam.
Game development is a high-risk, high-reward industry now. It’ll reward people who are creative, industrious, and lucky. The number of stable careers will decline in both quantity and pay, and capital owners will slowly pivot to buying out game IP instead of developing in-house. The only way to maintain dominance in a cutthroat, high-risk industry is to outsource the risk.
Consumers will win and producers will lose. Such is the nature of competition.
-
Comment on Coinbase says cost of recent cyber-attack could reach $400m in ~finance
TonesTones Link ParentAs far as I understand, the primary use cases for cryptocurrency are crime and speculation, and people usually don’t speculate with billions. I suspect the victims of most of the theft are...As far as I understand, the primary use cases for cryptocurrency are crime and speculation, and people usually don’t speculate with billions.
I suspect the victims of most of the theft are criminals themselves.
-
Comment on AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms in ~comp
TonesTones Link ParentReinforcement learning is a technique to train a model where you give it the ability to evaluate itself, and crucially, evaluate itself well, so that it can self-correct undesirable behavior....Reinforcement learning is a technique to train a model where you give it the ability to evaluate itself, and crucially, evaluate itself well, so that it can self-correct undesirable behavior.
Games are the obvious example, since a “win state” and “lose state” are respectively desirable and undesirable, so you can train the model to play games. I brought up chess engines versus chess reinforcement learning since the ability to roughly gauge the strength of a position accelerated the learning of chess models.
Reinforcement learning is common, but only where people have figured out how to use it, since it’s tricky to use well. LLMs cannot use reinforcement learning; they can’t automatically verify if their answer is correct or a hallucination. A human being has to label the answer as “good” or “bad” and then add it to the dataset (hence, supervised). The reality is a bit more nuanced; usually an LLM involves a bunch of models: some trained with RL, some supervised, some a mix.
But the overall supervised learning heuristic of “the quality of the dataset determines the quality of the model” still holds.
I honestly can’t put it better than Wikipedia (emphasis mine).
Reinforcement learning differs from supervised learning in not needing labelled input-output pairs to be presented, and in not needing sub-optimal actions to be explicitly corrected.
A supervised model can extrapolate outside of its dataset, but generally not very far. Reinforcement learning, when it works, allows models to find any technique that improves the objective goal, even if humans don’t know the technique. So, the model can make improvements to itself.
Right now, it’s quite difficult to give LLMs a way to evaluate themselves. However, this announcement shows a model that has done better than state-of-the-art (i.e., every human ever) on several open questions in math. And they describe the model as “evolutionary”. And they say that the model can only solve “verifiable” problems that can be described as an algorithm.
These are all hints of a model trained with reinforcement learning. Big picture: they told their model to find more efficient solutions to well-defined math problems. The model did better than any human. That scares me.
-
Comment on AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms in ~comp
TonesTones LinkEveryone has their take on LLMs and whether or not they are useful. The whispers in the AI research communities have always not really cared, because LLMs never were the holy grail. LLMs are...Everyone has their take on LLMs and whether or not they are useful.
The whispers in the AI research communities have always not really cared, because LLMs never were the holy grail. LLMs are supervised learning models, and will always be constrained by the power of the dataset.
The holy grail has always been language models that can train with reinforcement learning. LLMs are the equivalent of a chess engine for human language. They can look at a string of language tokens and tell you qualities of those tokens are reasonable.
Chess is amenable enough to engines that this approach can best humans; in language, LLMs can best most humans in most fields, but not humans in their specialized fields.
A reinforcement language model is like the AlphaZero of chess. Given an objective and a framework, it’s told to find a way to win. In every single game that we can design a reinforcement learning algorithm for, that approach dominates humans in ways that cannot really be understood by humans.
LLMs scared me because they represented a probabilistic framework by which to evaluate language, so that RL might be on the table. RL is really, really hard to pull off. Frankly, I hope this isn’t a model that can use LLMs as a framework to play language-based games to accomplish objectives, enabling optimization with RL. If it is, this is the holy grail of AI, all previous takes about GIGO are moot, and the age of humans being the dominant intelligence on this planet is over.
-
Comment on Why aren't Americans filling the manufacturing jobs we already have? in ~life
TonesTones Link ParentI don’t think this is how comparative advantage works. Goods in the U.S. are expensive, in part because of a high density of capital owners and high-skill laborers, and in part because of the...There will never be global parity in quality of life until there is a global minimum wage.
I don’t think this is how comparative advantage works.
Goods in the U.S. are expensive, in part because of a high density of capital owners and high-skill laborers, and in part because of the dollar’s status as the global reserve currency.
I’ve heard many economists say something along the lines of “laborers in countries with weaker economies and weaker currencies can get far more QOL benefit than U.S. workers for the same pay” when criticizing the president’s “plan” to bring back U.S. manufacturing. In other words, manufacturing will be cheaper in countries where workers are willing to work for less pay in raw dollars.
Unequal distribution of resources will always cause different places to have different strengths. This means that different wages will map to different qualities of life in different parts of the globe, probably forever. Thus, different parts of the globe will have different economic strengths and weaknesses (therefore, comparative advantage).
For example, cooks and service workers that live near coastal towns with nice weather will likely always make more money than the same careers anywhere else, but their cost of living will also be higher due to the relatively higher demand for living in that area. We can absolutely achieve global parity in quality of life without global parity in wages.
That will require additional investment and development in countries that don’t currently have a high quality of life.
-
Comment on Everyone is cheating their way through college in ~tech
TonesTones Link ParentI appreciate your optimism, but that requires acknowledging that the cheating problem was bad in the prior years. Right now, most administrations would prefer to pretend there is no problem....I appreciate your optimism, but that requires acknowledging that the cheating problem was bad in the prior years. Right now, most administrations would prefer to pretend there is no problem.
That’s why these articles keep popping up on Tildes; faculty are going to the press because internal feedback loops have stopped working, to force the hand of leadership.
Cheating is bad for Morale for the rest of the students.
Absolutely!! Worse, morale isn’t the only concern. If your administration begins ignoring cheating, you introduce an incentive. Internships, research positions, and post-graduation applications are competitive; if you ignore academic dishonesty, students might cheat to buy themselves more time to make other pieces of applications and experience stronger.
I don’t think this is actually beneficial; if your classwork is teaching you something, it’s probably valuable. However, some of the hypercompetitive programs do breed this kind of thinking; and if you are smart enough to both be able to verify an LLM’s output is correct and tweak it to not look like an LLM, it’s extraordinarily low-risk.
I don’t know if that’s happened yet, but it wouldn’t surprise me if it has, and it will happen eventually if we continue on the path at hand.
-
Comment on Everyone is cheating their way through college in ~tech
TonesTones LinkThis will be one of my most cynical takes on this website. There exist ways to catch student cheaters. Students using ChatGPT to generate assignments is absolutely not an unsolvable problem. It’s...This will be one of my most cynical takes on this website.
There exist ways to catch student cheaters. Students using ChatGPT to generate assignments is absolutely not an unsolvable problem. It’s a hard problem, perhaps the hardest problem educators have had to face, but not impossible.
At least, not impossible from a technical perspective.
If you look at college rankings, one of the factors they use to rank colleges are graduation rates: particularly, 4-year graduation rates. All the top universities pride themselves on having 80-90% of students graduate smoothly within 4 years.
If I had to guess, I’d estimate about half the students at most universities are regularly cheating. I hope this strikes most readers as absurdly high, but honestly, I’ve had multiple enrolled students at different universities give me estimates as high as 80%. I have a hard time believing that, but an estimate with such a high proportion tells you something about their priors.
The impossible problem isn’t catching cheating students. The impossible problem is how to deal with the issue without dooming your institution’s reputation. The first school to actually enforce a cheating problem would have two options:
A) Unjustly select a subset of cheaters to punish, and let the rest pass and graduate.
B) Deal with the public backlash of being the “cheating” institution.
I bet faculty are so frustrated with administration’s decisions because there’s a fundamental incompatibility between an educational institution’s mission and the reality of the situation at hand.
Better to pretend the problem doesn’t exist at all.
-
Comment on No, it’s not the incentives—it’s you in ~science
TonesTones Link ParentI absolutely agree that the way towards meaningful change is via systemic change. All the changes you’ve illustrated would make science a better place to work. I also agree that the author’s...I absolutely agree that the way towards meaningful change is via systemic change. All the changes you’ve illustrated would make science a better place to work.
I also agree that the author’s article isn’t really meaningful advocacy. I do think the author believes they are pushing for change when they really aren’t.
They spend a lot of time highlighting what can and does go wrong by working within the system, instead of addressing why the system exists the way it does.
Those words still deserve to be said. Human systems don’t exist in a vacuum; they are made up of their individual components: humans. Systems only change when the humans within them change, often against the incentives that be.
For example, if everyone in industry maximized their salary, non-profits and other good-doing groups would struggle to find employees. If regulators always capitulated to the demands of lobbyists, society would suffer.
The author is criticizing the mindset that the systemic incentives deprive humans of agency. They don’t. You still have many, many individual choices. The hardest choices are always those that go against the incentives, and they often make the biggest difference.
During my time in academia, I made a set of choices that went against the incentives that existed. They may have cost me my career within academia. I also believe those choices likely galvanized enough political will within my community to make small but meaningful systemic change. I wasn’t solely responsible——most of the credit for those changes should go to those in leadership positions——but I have a sense those changes wouldn’t have happened without me (otherwise they should have happened years prior).
I won’t argue that there is a moral imperative to make those extremely difficult decisions. I think it’s perfectly okay to behave as the incentives drive you to. I think it’s crucial to acknowledge that there is an opportunity cost to that behavior. There is always individual agency towards making the changes you do want to see.
-
Comment on Japan has successfully used drones to trigger and guide lightning strikes - and keep flying in ~tech
TonesTones Link ParentI really appreciate the ‘color me jaded’ pun and will be stealing that in the future. The drones rely on the ambient electrical field to work; they are directing the strike, not generating one. I...I really appreciate the ‘color me jaded’ pun and will be stealing that in the future.
The drones rely on the ambient electrical field to work; they are directing the strike, not generating one. I don’t think this counts as “optimistic”, but I imagine drones armed with munitions do about as much damage to infrastructure without actually requiring a thunderstorm in place.
-
Comment on Japan has successfully used drones to trigger and guide lightning strikes - and keep flying in ~tech
TonesTones LinkZeus would be furious. This technology is so incredibly cool. It’s the kind of thing you look at and wonder if you are living through science fiction. I wonder if someday, if made precise and...Zeus would be furious.
This technology is so incredibly cool. It’s the kind of thing you look at and wonder if you are living through science fiction. I wonder if someday, if made precise and predictable enough, this could be used for lightning shows if you are in a strong enough thunderstorm.
-
Comment on New study attributes nine trillion dollars of climate change related damages to just five companies, and outlines how they could be held accountable for specific local damages in ~enviro
TonesTones LinkInteresting read. I’ve only read the Perspective section, so forgive me if the paper addresses this. The method described simulating the climate supposing one company hadn’t emitted anything to...Interesting read.
I’ve only read the Perspective section, so forgive me if the paper addresses this. The method described simulating the climate supposing one company hadn’t emitted anything to capture damages, and notably, includes energy products sold by the company.I’m glad to see any progress towards addressing the issue of climate change, but I see a few glaring flaws with this approach.
(A) Even assuming the courts take the simulations as flawless, legal action here will only ever be brought against the largest emitter. You can run two different simulations: Simulation A and B if Emitter A’s and B’s emissions, respectively, are removed. If Emitter A has emitted more than Emitter B, the damages in Simulation A are always greater, and the legal framework described in the paper only allows legal action to be taken against one emitter without compromising both cases. (I think the claims “But for A’s actions, I would not have been harmed” and “But for B’s actions, I would not have been harmed” are logically incompatible.)
(B) I am not a lawyer, but including the products that were sold in the damages (like the gas sold to gas stations and eventually to car drivers) appears to leave a gaping hole in the legal argument. Specifically, you cannot assume if the emitter took no action that the damages would not have occurred. Any driver of a gas car would be able to testify “If Chevron didn’t sell gas, I would have bought gas somewhere else.” Again, you lose the “but for X’s action, I would not have been harmed” claim.
All this reading demonstrated to me is how legal system lack the preparation to handle the real damage of climate change. I’m never going to say they shouldn’t take the cases to court; “don’t let perfect be the enemy of the good”, and all that.
However, we need to recognize that even if climate change damages do not fit neatly into our legal system, action still must be taken. The legal system must fit society’s needs, not the other way around. In times of crisis, you need leadership. Sometimes, you need leadership that push the boundaries of the law. Conform too strongly to the status quo of your time and risk forever perpetuating the problems of the present into the future.
-
Comment on OpenAI is a systemic risk to the tech industry in ~tech
TonesTones LinkThere’s a theory I’ve heard floated from some pretty well-respected people in finance that this article backs up. I haven’t heard anyone endorse this theory, but they do talk about it. As I...- Exemplary
There’s a theory I’ve heard floated from some pretty well-respected people in finance that this article backs up. I haven’t heard anyone endorse this theory, but they do talk about it.
As I understand it, the idea requires these assumptions:
A) Competitive AI systems have opex on the same or greater order of magnitude as capex.
B) Tech leaders believe AI is a “sport of kings”: the AI system the captures the majority of users will take all the profits (ala search engines and social media).
C) Most users always choose whichever AI is free regardless of relative quality.So long as these three assumptions hold, there will always exist a tech leader willing to subsidize the opex of their AI system to capture the user base, and you can conclude that no AI company will ever win the sport, and every single company is going bankrupt.
In other words, if most of your debt is used to subsidize your opex, and a competitor will be willing to subsidize their opex to capture your user base, then you cannot convert to profitability without losing your users, and you must subsidize opex in perpetuity.
I think most investors believe either A or B fails. If A fails, the good AI systems will be so expensive to build that it precludes new competitors from entering the space (if this holds, then you only need to subsidize for longer than your competitors). If B fails, companies will find a way to be profitable without capturing a huge number of users. (This is why the belief is the important axiom, it informs the willingness of tech CEOs to subsidize their opex in the short-term.)
I do not know what OpenAI believes; they currently have the vast majority of users but they cannot remove their free plan without losing users. And they cannot subsidize longer than Google or Meta who can self-subsidize with their money printers.
I don’t know what I believe. But when you lose sight of unit economics, interesting games are bound to appear.
-
Comment on What programming/technical projects have you been working on? in ~comp
TonesTones LinkI’m always a little nervous posting here will cause people to deanonymize me but I think this project is broad enough. I’ve always envied those Linux power users that can boot up a new machine,...I’m always a little nervous posting here will cause people to deanonymize me but I think this project is broad enough.
I’ve always envied those Linux power users that can boot up a new machine, run a script, and have the operating system customized exactly to their preferences.
However, I’ve tried Linux as my working OS a few times and never could quite click with it; too many good GUI apps on MacOS that I like to daily drive.Recently, my computer broke and I needed a fresh install of MacOS. So, I’m building a script to do all of my MacOS customizations from the command line.
This project has been involved. Some sidequests include:
- Sketching a dependency tree to figure out what order my configurations need to happen in, since I start with just a Z shell with the default environment.
- Deciding on a semi-secure way to avoid many superuser authentication prompts without running the entire script as the superuser (for MacOS reasons).
- Integrating other objectives into this project (like dotfile version control).
- (In-progress) Building a small CLI tool to instantiate a bunch of the same environment variables in a shell-agnostic way.
- Learning my way around osascript and a couple other MacOS command line tools to do all the OS-specific customizations like system preferences.
- Trying to minimize my use of osascript by just putting app-specific configuration files into Application Support.
Normally, I’d get exhausted learning this much. However, this project has activated my sense of “holy shit, I’m a wizard” in a way no other project has.
Finishing a piece of the project, running it, and watching my OS just transform in front of me is super fun.I’ve already committed to show my work to a friend by factory resetting my computer when I’m finished. The stakes are high.
-
Comment on The dangers of vibe coding in ~tech
TonesTones Link ParentThe Jevons Paradox, which I learned about somewhere else on Tildes, is possibly relevant here.Throughout computer science history, there have been many advancements that improve the productivity of programmers. Every single one has paradoxically increased the demand for programmers, not decreased it.
The Jevons Paradox, which I learned about somewhere else on Tildes, is possibly relevant here.
The Trump administration strikes their newest blow in their war against higher education.
I don’t know about the legality of this decision, but it seems illegal to me.
Something tells me the students will be deported regardless of what the courts decide.