xyzzy123

7 ms·

61.
by xyzzy123 5mo ago
Weird question, do you think AIs might prove a lot of theorems that are mainly useful to other AIs (i.e, make nearly no impact on the human culture of working mathematicians), which then get used to prove results that humans do actually c
62.
by xyzzy123 5mo ago
Funnily enough a lot of this "extension sprawl" is caused by the _difficulty_ of installing tools on locked down Windows machines. I recently moved to a locked down SoE and instead of being able to use regular tools (which require
63.
by xyzzy123 5mo ago
Also predictability, resilience, sovereignty. I'm not worried about other people's outages, that unexpected demand will impact me at an inconvenient time, that someone's watering down my model, that my costs will change unpre
64.
by xyzzy123 5mo ago
Right but at n=1 you are writing the policies, reviewing them, signing off on them yourself, raising risks, then ducking out the back and running out again with a fake mustache so you can accept the risks you raised... regarding yourself...
65.
by xyzzy123 5mo ago
If you get someone else to do it you are not responsible for failure.
66.
by xyzzy123 5mo ago
They didn't mention what you do with the microplastic goop after you empty the filter. Landfill?
67.
by xyzzy123 5mo ago
I think the idea is that you can launder your team or product AI spend through your AWS account. This matters in Enterprise. It looks like the difference with Bedrock is that you access more "Claude platform" stuff than just the m
68.
by xyzzy123 5mo ago
I had dup uuids causing soak test failures in a Linux based distributed system. After long investigation it turned out there was a kernel bug (race condition) that meant two processes on MP system reading from /dev/random at the s
69.
by xyzzy123 6mo ago
I don't see that it makes much difference until we know the distribution of issues that Mythos finds and how reliably it discovers them? Vulns from inspection are discovered via a stochastic process of someone looking at the code, know
70.
by xyzzy123 6mo ago
The concept totally works but I would worry about using a beelink in a business context where I had to support it. For up to low hundreds of users I think you're better off just with 1 vertically scalable box for all the officey /
71.
by xyzzy123 6mo ago
Personally I don't think coding agents will regress significantly as long as there is competitive pressure and independent benchmarks. Regulation is a risk because coding may be equivalent to general reasoning, and that might be limit
72.
by xyzzy123 6mo ago
I mostly agree with your experience, but; Every day I deal with bad judgement calls from humans (sometimes my own!), but I don't screenshot them because it's not polite. I don't think we're at the top of the curve yet? C
73.
by xyzzy123 6mo ago
ReDoS at CVSS 8+ ... in the configuration file parsing of a bundler.
74.
by xyzzy123 6mo ago
Kernel headers out of date -> kernel vulnerability... in a container.
75.
by xyzzy123 6mo ago
I think the novel idea here is you jam some hardware together (whatever you like) that can do "physical real world" things with a well understood interface and then spin up Claude with access to it. The way I'm thinking about
76.
by xyzzy123 6mo ago
Is it the industry making this choice or the customer? You could make a car that's safer than others at 10x the price but what would the demand look like at that price? Would you pay 2x for your favourite software and forego some of th
77.
by xyzzy123 6mo ago
I dunno if Steve was autistic or no but there's a subtype of people who develop a "special interest" in reading, influencing and manipulating people and get quite good at it. They study people carefully - each encounter is a
78.
by xyzzy123 6mo ago
I just find that e.g. cli tools scale naturally from tiny use cases (view 1 ticket) to big use cases (view 1000 tickets) and I don't have to have 2 ways of doing things. Where I DO see MCPs getting actual use is when the auth story for
79.
by xyzzy123 6mo ago
Hmm, but you can't write a standard MCP (e.g. batch_tool_call) that calls other MCPs because the protocol doesn't give you a way to know what other MCPs are loaded in the runtime with you or any means to call them? Or have I got t
80.
by xyzzy123 6mo ago
My main complaint with mcp is that it doesn't compose well with other tools or code. Like if I want to pull 1000 jira tickets and do some custom analysis I can do that with cli or api just fine, but not mcp.
81.
by xyzzy123 6mo ago
The DCs are in VA because its fed / spook central, which induces a large number of well paid policy / compliance / paperwork / tech jobs etc. NoVa literally one of the richest regions on the planet and it's all sort
82.
by xyzzy123 7mo ago
I'm not claiming the US is "good"! I'm saying I would prefer a world where IRGC doesn't have long range nuclear weapons.
83.
by xyzzy123 7mo ago
Conversely I find it strange how many people seem to want US servicemen to come to harm because it would validate their political views.
84.
by xyzzy123 7mo ago
We're not really calorie constrained anymore and most humans live in much denser environments than they used to. You would expect rate of exposure, the rate of mutation / change and the rate at which new pathogens appear to be hig
85.
by xyzzy123 7mo ago
I don't really see why most existing equipment would be usable in this way. When you automate a thing you often have to rethink the entire problem. But more generally, automation is for _repeatable_ things and a lot of research is... n
86.
by xyzzy123 7mo ago
Think for a minute about why they worked in shacks. Also worth noting that Birren was a paid consultant for DuPont - the company that made the paint. From that perspective alone you would kinda expect he is gonna pick colours they already p
87.
by xyzzy123 7mo ago
Great discussion on colour theory but I think it skips over the fact that chemistry and economics were also real constraints. Paints that don't fade easily, chemically inert, durable etc and can be produced in enormous volume are not t
88.
by xyzzy123 7mo ago
I'm super confused. The small model "cost field" `rag-api/geometric_lens/cost_field.py` was trained on PASS_TASKS like "Write a function that counts vowels in a string." and FAIL_TASKS like "Write a f
89.
by xyzzy123 7mo ago
Having shell is extremely handy for further discovery. SO handy that if they were just gonna patch the bug and lock you out, you would simply not disclose it.
90.
by xyzzy123 7mo ago
Very often kids are un-knowing participants in status signalling by adults. Parents want to signal "this child is looked after and we have a lot of capacity". Clothes, lunches, a lot of things are quietly like this. A good hand-ma

More ›