The way this is done in every domain where the real is a bottle neck is with simulation. So the actual argument is: What domains do we think are not practical to simulate even with extreme intelligence making the simulation and the ability to bring immense amounts of computing power to bear while running.
I don't think there are actually that many domains where you can't simulate them with enough knowledge and compute. So then the bottleneck becomes your ability to sample the world and refine your models. For some domains you can have the models run against each other in your sim which greatly reduces the need for real world feedback (see AlphaZero for example).
So in the case of human persuasion, like asking a boss for a raise, the model could of failed many times a second while trying to persuade a simulation of a typical middle manager with the same parameters as your boss.
That's a very "Spherical cow in a vacuum" approach. You would need to accurately model the boss, as well as all your interactions with the boss up until your raise request, as well as the boss' environment when you make your request.
You could be on the wrong side of a reasonable boss, or in his good graces but during a bad day ...
That's the point the writer was making when he mentioned role playing games. In isolation, you could spend hours drafting the perfect plan, that wouldn't survive two minutes of contact with reality. You just don't know if your model is correct until you try.
> I don't think there are actually that many domains where you can't simulate them with enough knowledge and compute.
The problem is that you have to simulate them accurately enough and fast enough. A perfect simulation of physics will necessarily be orders of magnitude slower than physics. To accelerate it, you need to figure out which parts can be cut or approximated without derailing the model, and just that part will require a lot of back and forth with reality.
Success isn't guaranteed either. If a physical system does whatever it does optimally enough, fast simulation is just physically impossible. Furthermore, even simple chaotic systems require unrealistic precision in the initial parameters to be useful. Natural entities are fairly chaotic and also pretty darn efficient at what they do, so it's plausible that even the best possible simulation of a middle manager is a slower bottleneck than a real middle manager.
But how do you source the observations used that would be the basis of your simulation? Or are you assuming the real world won't change? At least in my mind I don't see many domains that remain static enough where continuous sampling of the real world will not be required. Go and Chess are artificially stable domains, the rules are fixed.
Any social domain where there's competition of resources (salary, funding) will definitely not be stable, it would require constant testing against the real world.
Ok, but this just means that model predictive control won and you're not modeling just individual concerns but rather, you're modeling the entire world.
Yes that is generally what you get down to: you can’t really own land free and clear in a populated area. You depend on the government in many ways to enforce property rights, as well provide services for that property and in return you have to pay for that.
It makes me uncomfortable too actually but I can’t see any alternative that does not get back to a similar spot or violence. In addition if we shifted to a land value tax, it would push taxes onto the very dense developed areas encouraging better use.
Do you have a citation for the horse number? My research [0] shows we have a 1/5 out peak number in ~1900 and our population has 5x since then. This means per capita horses in the country is down 96%, while absolute count is down 80%. Both are huge numbers.
Good call! They have info on general population which shows about double the number of horses I originally thought (at least in 2016), and declining at a good clip of 10%+/decade)
. So drop is still large but just a bit smaller.
The general point is before cars horses were used a lot, entire industries were based around them and our now gone. The toy factor does not compensate because of cost and inconvenience of owning a horse.
This missing bit is that many people are pain free with the same scans as you have. Surgery does help people, but it’s clear that the signals we use to say surgery is needed are very noisy and not as accurate as the surgeons would have you think.
Any details about the harness the agents were given? I am curious what representation of the screen and world state was provided to the agents and what tools they had available.
Oh sorry I should be more clear on that. Will add to report.
For agent harness I did Claude Code, Codex, Grok Build. This was primarily a cost driven decision — I have a lot of free tokens and I didn't want to pay API prices for this.
For game harness I used minimal BW-API issue command and get observation apis as tools. I felt this was the most fair way to do it on my small scale.
In the future I would like to integrate code mode and multiple games/I think if it was a best of 5 where each agent could learn from its past games and build its own automations over time that would be much more interesting.
Given that a lot of their failures are from fairly basic mistakes related to the unique setup (eg, thinking rather than defending immediately) I'd love to know how much they improve with basic tips.
Or possibly even whether they can learn from a game themselves. "Analyse your game for your failures" -> Then give a fresh agent of the same model that "learnings" doc for the next match. Do the rankings change over time, if models can write instructions for future selves?
10s of miles is close enough that they are functionally in the same place in a military context though. The latest cheap massive wave attack drones fly hundreds of miles and hour which puts the DCs less than a minute apart by flight time. This means if you want protection against those risks you have to use multiple regions instead of, or in addition to multiple AZs.
The headline here is off, PS5 Linux relies on bugs for installs, and the project couldn’t keep the last know hypervisor bug secret:
> Slop kiddies found the only hypervisor bug left, which I had also found a while ago, and decided to report to Sony,” the modder said on social media. “I asked them to at least wait for GTA 6 to come out so that people would have the opportunity to legally purchase the game and also enjoy linux. They agreed to wait, but not a day passed and they decided to waste it instead.”
So what has really happened is the LLMs have lowered the participation floor for this space enough the dynamics are changing. The new comers would rather have a few dollars vs. the founders who would rather have a project. It’s likely the projects days were numbered either way because Sony could also just take a frontier LLM and examine any released installer for the project and reverse engineer and bug.
Everyone in this scene knows upgrading your console is sure path to getting yourself locked out of current and probably future exploits. I don't think anything really changed here. The moment these exploits go public, they'll be patched in the next version.
This vulnerability will still exist in older versions. The project can still continue development of Linux on PS5 and, eventually, another exploit will be found that enables the installation on a later version.
Frankly, based on the language used, a resignation over this seems to have a lot more to do with the person being upset that LLMs are lowering the bar for entry. They seem more upset that their skills aren't as useful as they were.
It seems like the reputation as a PS5 hacker was the objective, not getting Linux running on PS5.
I don't think there are actually that many domains where you can't simulate them with enough knowledge and compute. So then the bottleneck becomes your ability to sample the world and refine your models. For some domains you can have the models run against each other in your sim which greatly reduces the need for real world feedback (see AlphaZero for example).
So in the case of human persuasion, like asking a boss for a raise, the model could of failed many times a second while trying to persuade a simulation of a typical middle manager with the same parameters as your boss.