Generating big chunks of code is rarely what I want from an agent. They really shine for stuff like combing through logs or scanning dozens of source files to explain a test failure. Which benchmark covers that? I want the debugging benchmark that tests mastery of build systems, CLIs, etc.
Probably want to look at SWE bench pro or terminal bench 2. They cover these longer horizon tasks that need more than just writing a bit of code in one file. And SWE bench pro in particular it is not yet saturated like many other common benchmarks. Normal SWE and LCB are not really useful anymore because they are already being gamed hard so the developers can quote high numbers in a repo readme or press release.
I've been speaking French since pre-school (albeit in North America mostly) and to me é always sounds more like the English short i (as in "tip"). I'm becoming increasingly convinced that everybody on Earth but me is wrong about it.
Do you happen to be from the western US or Canada? They tend to lower the /ɪ/ monophthong (i of tip, pit, sit, etc.) there, making it sound pretty close to /e/ (French é, German eh). It's one of those things that, combined with regionalisms and other accent features, give away where you grew up :) I noticed a lot of Londoners do this too, though this is just my experience.
I will do the work for them (typically paid for by my employer) iff I can expect them to fix it.
Blackbox debugging is a PITA, which is part of why I prefer open source, but it is what it is... If something is broken, and I can get it fixed by putting in the time to get a good report, and etc and they fix the thing, then I'll do it.
But if they don't fix the stuff, I have no shortage of things to fix myself.
It's not about some American cultural attachment to credit cards. It's another classic Apple frustration move where they make the experience worse for their users in the hopes they'll blame someone else like the UK govt. They do the same thing with green bubbles.
A small company can use a PEO like Rippling (a YC company) where employees are “co employed” with the actual company and for taxes/HR/benefits with for Rippling. It’s not like contracting. Everyone from the CEO down is “co employed”
I’m never going to defend the American health care system as I sit in a country now for six weeks where I fully plan to become a resident of post retirement mainly because of the healthcare even if I don’t live here full time.
My guy, congress can't even remove valid bad actors who openly lie to and threaten them. They will never fix any problems except their "light" pocket book
For a lot of people on HN, they grew up in VHCOL places that used to be affordable and are trying to keep the lives they've had since birth.
I was forced to move, lost connection to my friends and half my family, all the places I knew and had memories/attachment to, habits/hobbies. I understand why they are fighting to keep their lives and not give up and in a way die and start a new, lonelier, much different life.
I don't understand how that is nepo/spoiled/rich behavior, it's just basic normal human behavior. Thinking it's totally cool to displace people is also normal human behavior, just to me the shittier less justifiable of the two.
I agree with all of that, but it's not representative of everybody that grew up in the VHCOL area. A slight majority of residents in the Bay Area (used as an example) own their own home. If you grew up there, and your parents owned their own home, your family has benefited enormously from the meteoric rise in house prices. Those people (long-time residents) are thrilled by the influx of tech cash and actively pursue NIMBY policies to restrict the housing supply to keep prices as high as possible. Most of the tech workers actually moving to the Bay Area and renting would much prefer a massive increase in the housing supply to bring prices down.
California is an especially egregious example because none of the inherited familial homes are taxed appropriately, which lowers liquidity and drives up market rates further. If you wanted to create a landed gentry, California Article XIII A is the gold standard for a policy to do that [1]
Of course, a lot of families never end up owning a home in an area that will experience that kind of appreciation. But the idea that it's "newcomers vs. life-long-residents" is wrong. It's actually more about the tension between the life-long-residents who own property and pursue NIMBYism vs. everyone else.
My example is representative of EVERY person I grew up with that didn't come from generational wealth. I guess if their parents died when they were young of a fluke you would consider them lucky, but what's the average lifespan for someone in the area? Everyone I knew would rather have had the option to live/raise a family in their home town than inherit a million dollar home in their 50s after they had to start a new life they didn't pick.
You can write paragraphs about how displacing people is fair, how kicking grannies out of their homes and auctioning them off because of tax debt (something that was happening) is the moral way. But you are still just talking around displacement of people to reach your desired end goal.
A functioning economy is full of these tensions between people with divergent "desired end goals". Everybody wants high home prices when they want to sell but low home prices when their kids want to buy. Everybody wants low prices for things but high wages for people, even though those things are inversely correlated. I'd bet many of the parents you're talking about voted for NIMBY policies and cheered the tech industry's rise. Of course they would. If I'd owned a house in the bay, I'd have been pretty jazzed about it too.
I'm not pro-displacement. I'm pro-housing, which we need much more of in SF.
Those used to be quite reasonable cost of living areas. We're not talking about owning a mansion in the Hamptons, but a decent-sized apartment in a downtown area or a nice single family home in a pleasant neighborhood.
Until I see median real income start to actually go down, I just don't buy it.
AI is currently a commodity. Maybe one of the labs will be able to differentiate sufficiently to be able to charge the kinds of premiums they need just to pay back their investors. Maybe, instead, we'll see something akin to the FOSS revolution, where large, high-quality, open training sets are developed to make sure there's always a fair alternative to the big players. Then who actually benefits from AI? Mainly users, not companies.
In many ways, the bar to having a competitive advantage is actually lowering. I reckon in the future, simply avoiding a crippling social media addiction that sucks up 4-8 hours of every day will be enough to get rich.
That's what the "real" in "median real income" means. It measures how much stuff you can buy, not how much currency you have. And it's been going up relatively reliably: https://fred.stlouisfed.org/series/mepainusa672n
Median is simply a 50/50 centrality measure. When you're dealing with a bimodal distribution you won't learn too much about it with median. So you need something else, e.g. Gini coefficient to characterize it.
As for the real, does it measure housing, insurance and medical as well? Does it measure the safety on the streets? Overall quality of life?
AI will strip what's left of the middle class to bare bones. And the tech lords will call it 'disruption', the fixated adolescents they are.
And the reason for the death of the middle class is regulation that was purportedly to help reduce inequality, like all manner of taxes. I think the solution is obviously to keep trying. That's not the definition of insanity, that's good old fashioned human stupidity, and unlike AI stupidity, human stupidity gets reset back to 100% with every generation. No wonder history repeats itself.
Actually, median is exactly what you want. It strips out outliers. It's the "middle of the pack" person. Mean is the one that would be skewed by outliers.
And "the poor are getting poorer" is simply untrue for the last 10 years. They had a pretty bad time from 1980 - 2015, but in the last 10 years, their real income has risen faster than any other quintile: https://www.visualcapitalist.com/growth-in-real-wages-over-t...
Median says very little about distribution and says almost nothing about how the tails are doing (which are real people that are easily ignored).
That page breaks about a second after loading. It's enough time to see the graphic, but not enough to see the methodology for data collection. Can you share how that data is collected? Afaik government sources do not track real income distribution.
Yes, the bottom quintile has a lot more variance, but the argument that the poor are being left behind has been really hard to make for the last 5 years. Coupled with high inflation, which erodes the savings of the rich, they're actually getting a bigger slice of the overall pie.
You're looking at what percent of the total wealth pie do the poor get. But the pie itself is growing, and so is _everyones_ slice of the pie.
Maybe you think its an inherent problem that some people get a bigger percent of the pie than others. But its objectively untrue to say that the poor are getting poorer.
After a decade of "quantitative easing" there is still a shitload of money looking for a spot in the economy that needs it. NFTs, bitcoin mining farms, AI data centers, FAANG stocks, real estate, gold ... These are hedges (or attempts to hedge), but they add little to no intrinsic value to the world's prosperity.
They merely shift and concentrate wealth and power, for the most part.
> Until I see median real income start to actually go down
I'm not sure I understand this, it doesn't feel like what I have "lived" for the least 30 years.
Median real income might not be down statistically, but the purchasing power of professional incomes relative to housing, education, and major life costs clearly feels lower than it did in the mid 90s. An inflation-adjusted six-figure salary today does not deliver the same lifestyle position it once did.
Man... healthcare costs, too. Hell, even computers! Raw computing power per dollar is cheaper than ever, but the minimum spec required to function professionally has risen so much that the real cost of staying technologically current feels higher.
I had my first house built in the burbs of Atlanta - 2700 square feet 3-2 and a bonus room for $170k.
Going by the house shouldn’t be more than 3.5x your income. That puts the necessary income less than $50K.
Heck I had my second house built in the northern burbs of Atlanta in “the good school system” for $335k in 2016. We sold it in 2024 for exactly twice the price.
Sure but you have to run AI on something and today that's basically limited to multi-thosand dollar hardware - the price and effort to entry has to come down to make it make sense for normal consumers to move it out of the cloud to local.
I'm with you. If we all have access to AI then how is that a bridge being taken away?
Doesn't that mean that a single person can more easily disrupt the status quo?
All this stuff about genetics... I just don't think it's relevant at this point. Average intelligence and access to the internet is what most of the world has.
It's the systems of money and law that are taking the bridge away not AI. But someone could invent new systems to replace the ones that don't serve the 99%
Will most people go that far? Probably not. But the bridge is still there - unless they take the AI models away entirely.
I think the only way the rich can stay rich with ai is if they just use AI to convince people that they can't do anything themselves. After all that's what the last century was about with respect to capitalism.
I think my concern would be if the relationship between model intelligence and inference cost was altered very significantly. I sort of feel like we got lucky that AI isn't arbitrarily scalable in a single instance
(i.e. if you could run a single LLM on an entire datacenter and it just immediately becomes a super genius versus running it on the minimum viable hardware i.e. some form of quantization on a local machine.)
Obviously there's a sort of goldilocks zone / most appropriate substrate for an LLM to run on somewhere in between those two extremes (small cluster of tightly coupled flagship GPUs)
So luckily enough the economics appear to work out to make that at least conceptually viable for even private members of the public to afford access to the same order of magnitude of LLM intelligence. But we're already seeing some departure from that.
My concern would be if this curve was altered significantly by a new algorithmic approach beyond or instead of Transformerd such that someone with $200,000 to spare could achieve just like a completely categorically different quality of work, massively magnify their existing wealth advantage, because this would be a threat of the sort being discussed above, namely a pathway to a severe form of modern Feudalism.
Well, I think in the context of the parent comment, separating out housing would risk overstating changes in its effect on purchasing power because increases to housing would already be captured by inflation (since
we're talking about real median income, which is already inflation adjusted)
I agree that housing affordability is a major problem and that looking at it independently could help you quantify if housing specifically has become more unaffordable, but that's a different question then whether the median person's overall purchasing power has declined (considering all of housing, healthcare food etc)
Yes housing prices are captured by inflation, but since housing prices are so different from region to region and city to city, just looking at a country-wide average median income does not say much.
housing prices and health care are captured by the "real" part of "median real income". Inequality is not, but as long as everybody is getting richer, I'm less concerned about inequality.
Inequality is still important even if everybody is getting richer (I disagree that everybody is getting richer, but that's a different discussion). Because wealth and power concentrated in the hand of a few means that a few have outsized influence on politics and society, which is very much eroding the fundaments of how a democracy is supposed to work.
Furthermore, there are many studies showing that more wealth inequality results in more consumption in a society. Which is not good for many reasons.
> Until I see median real income start to actually go down, I just don't buy it.
Assuming you're intending "real" to mean the technical definition of "real" which is "adjusted for inflation", its basically been flat since 2019*, and that's using the government's inflation measures which abuse things like basket substitution and other hacks to hide the actual increases in the true cost of living. If you made better assumptions about inflation, you actually would see that median real income is down dramatically already over the past several decades.
Your assertion that it's "flat" is rather unconvincing after looking at that graph. It's clearly been rising relatively consistently. The dip for covid I think is especially understandable, and it's currently higher than it's ever been.
44% housing, 8% medicine, 14% food. I don't see how these number are manipulated to the point of malicious data picking. They also publish their methodology.
It dipped during the covid recession and then recovered to an all time high. Is this your first time looking at an economic chart? It doesn't need to increase every quarter for it to have a very strongly increasing trend.
If you actually look at the rest of the graph, you'd quickly notice that 2 year dips are entirely normal. The overall trend is consistent. And it's risen over 5% in the last 4 years. Those are the 4 years since ChatGPT.
reply