M3 module was formalized fully purely from experimental data and from a nudge by earlier versions of codex in 15-30 minutes in a simple write/compile/fix-first-error loop. I was a bit surprised how fast it picked up the pattern but given there was a paper from '70s it became clear why later.
He did mention in one of the planet earth 2/3 series how so much of the land is used for farmed animals.
And, for the sake of completeness of argument, for restoring what was lost, the challenge is how to raise the standard of living fast enough for people so they give a damn about anything apart from ourselves was THE challenge to combat climate change and global ecological disaster. He specifically mentioned e.g. educating girls and making older-aged societies more propsperous. Prosperous people can make better choices about farmed animals as food.
Extremely insightful re: capacity to care for others.. though the trend of "NIMBY" and "petite-bougie" ideology (a more modern term being "temporarily embarrassed billionaires") and pretty much the guiltless profit maximization of the oligarchy point to that hypothesis being another form of trickle down economics (only the "prosperous" part. the educating women part should be a given, but if you read the comments in a recent frontpage post about humanitarian aide, the findings of those with decades in direct involvement pointing to "gender empowerment" and road infrastructure as most impactful long-term were met with the same "pronoun" schtick.
Modern agriculture, both animal and non-animal versions, are bad for the environment. Artificial fertilizers, replacing forests with farm land, and drainage of wet lands are all heavily contributing to emissions and water pollution, destroying local ecosystems as well as warming the planet. Artificial fertilizers is particular bad since its production uses fossil fuels, has large amount of accidental green house emissions, and causes eutrofiering to the point of areas like the baltic sea becoming basically dead from loss of oxygen. Runoff from farms are also now the primary cause of ecosystem collapse in fresh water lakes.
The most impactful elements of modern agriculture are entirely animal-based. Full-stop.
You in fact rightfully but incompletely recognize : artificial fertilizers (for giant mono-crop fields of soybeans to feed to cows and pigs [0]), replacing forests (to clear room for soybean fields and pasture for cows and pigs [1][2]), and runoff of these fertilizers and manure into waterways. The parent comment is right - if we want to fix these problems, we must stop killing and eating animals at such an industrial and horrendous scale.
If it was a competition in who did more harm the distinction may be relevant but in terms of saving the environment and turning back the ecosystem back to a sustainable one, the distinction between animal based or non-animal based is mostly irrelevant. Modern agriculture are not sustainable for the environment. The fossil fuels that are pulled from the earth and put on fields are not sustainable, and the amount of run off that goes into the water are destroying ecosystems with no time table if they ever can recover. When different species goes extinct they stay extinct, and the distinction that "well, its not as harmful as animal-based agriculture" will not bring them back. The Baltic Sea an loud warning signal of what happens if we continue to go down this path of modern agriculture.
One of the few areas of sustainable farming is aquaculture like shellfish and seaweed, which could actually be used to reduce the negative effects caused by modern farming. If there were a competition in least amount of harm, those would likely be the winners.
Fields of corn or soybeans will still exist without animal-based agriculture, especially with current demand for biofuels. As long as the land can be farmed to generate revenue, people will farm it. Artificial fertilizers is the primary enabler of this.
No one is saying fields of corn and soybean wouldn't exist, but we would have far less of them without animal agriculture. You are creating a false dichotomy of "sustainable"/"not sustainable" the reality is human societies would be much more sustainable without animal agriculture. More sustainable does in fact result in less extinctions/ecosystem impact and reduced climate change.
Maybe I'm wrong but reading your comment it feels like you are letting the perfect be the enemy of the good, and you use your conclusion that we will never be sustainable as your excuse to continue to eat animals.
> No one is saying fields of corn and soybean wouldn't exist, but we would have far less of them without animal agriculture
There is no evidence that there would be far less farm fields without that. Farm fields exists if there is profit to be had. Right now the demand for biofuels are directly competing with the demand for animal feed. Farmers will primary grow and sell crops based on what pay the most, and can easily switch if one pays more than the other.
Notice that none of those says that farmers would not use the fields if the current most price worthy crop would go away. Farmers choose what to farm based, among other things, the market. If you remove animal agriculture, you don't get far less fields. You get fields with a different crop in them.
The only thing that will stop farming is either if the external cost of farming is applied, such as pollution, or if climate change makes farming the land unprofitable. Currently that pollution is not applied as a cost. A carbon and water pollution tax could be a strategy that addressed this, and would impact all farming regardless of crop. If that is "perfection" and "enemy of the good", then the definition of perfection is not shared.
> If you remove animal agriculture, you don't get far less fields. You get fields with a different crop in them.
we just simply cannot know this. but, we do know the impact of our current practices, not to mention the (supposedly) debatable ethical element of killing hundreds of billions of animals each year.
> in terms of saving the environment and turning back the ecosystem back to a sustainable one, the distinction between animal based or non-animal based is mostly irrelevant
this is simply false - did you follow any of the citations? you’re welcome to find something to support your position but as they say: if it can be asserted without evidence, it can be dismissed without evidence.
strongly disagree; we can quantify the differences. our current population levels are sustainable if we stopped overfishing our oceans and raising billions of animals each year for slaughter. check my other comment for a bunch of citations!
Regret analysis in bandit and similar algorithms shows how inference is connected to loss function. If your loss function is good, greedy inference is as good as joint inference.
Training on cost-to-go loss is good enough.
Perfect cost-to-go eliminates the need for global algorithms and allows local decision making. Given “natural” datasets it is probably the best thing to attempt to learn.
The fact that probabilistic graphical models never really worked proves it somewhat.
Are there any good papers on this you would suggest/specific search terms?
I am vaguely aware of some stuff, but would love to study more, I don't quite understand what this is all about (but I do see how LLMs can do attention to all prior tokens so you don't have the single-point-of-failure HMMs do which more necessitates Viterbi decodes)
Github was already struggling with bazillions of throw-as-much-crap-on-the-wall software running in actions, and now the world is running throw-as-much-LLM-crap-on-the-wall computation, as unstoppable as the pre-LLM era. Turning compute into excrement as fast as the planet is filled with it. Excrement being "Github Copilot code review" in compute world, and no need to draw what it is in our real world.
Weird that Anthropic decided to build a Claude Code Routines toilet.
If they aren't "smart enough" to know if it work they most likely are also unable to verify if the Lean formalization is indeed the one that matches the problem they were trying to solve.
Verifying that every step in a (potentially long) proof is sound can of course be much, much harder than verifying that a definition is correct. That's kind of the whole point.
That's not what the parent comment meant. They meant checking the Lean-language definitions actually match the mathematical English ones, and that the Lean theorems match the ones in the paper. If that's true then you don't actually need to check the proofs. But you absolutely need to check the definitions, and you can't really do that without sufficient mathematical maturity.
Yes, and the child comment’s point is that formalizing the problem is likely easier than having the LLM verify that each step of a long deduction is correct, which is why Lean might be helpful.
But both of you are ignoring the parent comment! Actually you're ignoring the context of the thread.
Originally someone said "I wish I was math smart to know if [this vibe-mathematics proof] worked or not." They did NOT say "I'd like to check but I am too lazy." Suggesting "ask it to formalize it in Lean" is useless if you're not mathematically mature enough to understand the proof, since that means you're not mathematically mature enough to understand how to formalize the problem.
Then "likely easier" is a moot point. A Lean program you're not knowledgeable enough to sanity-check is precisely as useless as a math proof you're not knowledgeable enough to read.
It’s not useless, because you can, for example, ask multiple frontier models to do the formalization and see if they agree. And if they have surface-level differences in formalization, you can also ask them whether apparently-different definitions are equivalent.
This isn’t perfect of course - perhaps every single model is wrong. But you are too quick to declare that something isn’t useful for arriving at an answer. Reducing the surface area of what needs to be checked is good regardless.
That's great if it works. But it's way harder to produce a formal proof. So my expectation is that this will fail for most difficult problems, even when the non-formal proof is correct.
I can't find it now but I once read that people who report the "paradoxical reaction" to stimulants have a significantly better outcome from stimulant medications, and both those values seem to have higher heritability than ADHD as a whole, possibly even linked to a single gene.
Following the bottle-cap madness, I don't think any current data shows the actual issue was resolved. Even worse, the effect on marine life is still not measured, and afaik reduction of harm was the primary goal. Instead of brutally high fines on fishing net waste, we got bottle-cap madness.
We have so much experience with scientific method, yet these massive decisions are adhoc, that's how the whole world works. We never tested what would happen by allowing mass production of plastic, or phones, or whatever, so these antipatches are going by the "feels" as well, with no individual taking responsibility for failures.
There is monitoring of beach pollution but data publication is typically delayed, like, we know it went down by 30% from 2015-2016 to 2020-2021, we will have data on a regulation that went into force in late 2024 only in a few years.
A bit off topic, but I recently had a drink that didn't have the attached bottle cap while travelling abroad, and my god, was it a minor annoyance to have to hold the bottle cap in my hand while drinking. I also almost dropped it because I expected it to stay attached. Funny how fast we adapt.
haha, all of a sudden I see a tab "waifu pillow" on Amazon, and think I have a split personality that runs searches in between consciousness shifts, and then I come back to a funny message.
https://github.com/vjeranc/fixed-rtrt
M3 module was formalized fully purely from experimental data and from a nudge by earlier versions of codex in 15-30 minutes in a simple write/compile/fix-first-error loop. I was a bit surprised how fast it picked up the pattern but given there was a paper from '70s it became clear why later.