Why would one need a blackboard if one had a coordinating agent that would launch a sequence of builder agents that run either in parallel or in sequence depending on what the coordinating agent determined was safe?()…
Translation: As the number of bugs per line of code is a constant depending on language and project, we will experience a world full of bugs now that LLMs generate so much more code. On the other hand, LLMs are so quick…
I noticed back in the days that my FidoNet point client software ran a lot faster (actually processing a lot of messages, so IO and CPU heavy) in Windows than in DOS, which was unintuitive to me back then.
What was the minimum required density you derived?
Not if the models are way cheaper. For many tasks you can do with DeepSeek Flash, and for more gnarly problems you switch to GLM or Kimi. But sure, if you do everything in large models like Kimi K3 or GLM it gets…
Until the manufacturers enter an arms race and copy a page out of the mobile hardware vendors book. But perhaps they'll do it badly and we've got a few more years.
Apparently the task was to remove dead weight. Haven't read what they replaced "Enzyme" with. I did subagent based removal tasks a few times. These were the ones that required the least amount of input or thinking from…
Taste is something you use to build the guardrails of AI over time and many paper cuts endured. Once you have a big enough set of these (call them rules) recorded, they become your development substrate. You narrow the…
Shameless plug for my project, which is a bit different. I had the problem of crashing terminals due to OOM kills and wanted to switch to a more lightweight terminal. Also I had too many of them and couldn't find the…
OS is listed as FreeRTOS.
https://en.wikipedia.org/wiki/Lean_(proof_assistant) Not sure about "bugs" in that area, but there is a lot of work going on by mathematicians in formalizing and checking ever more complex proofs using proof assistants.…
And reading reports, endlessly.
If these proofs were output by the AI in a format readable by a proof verification system, the verification step of publishing vanishes. Then it's only valuable to check if the stated intention actually matches the…
Then it's a case of trying to uphold the narrative?
Tried it on Android and got "!!!!!!!!!!!!!" for answers.
Fair take if the user is the same person as the implementer. If I sell software to my customers that reduces their operational complexity by me investing some code, then I don't consider it a loss.
Can be that I remembered it incorrectly, as it was in 2008, but here it is: https://se-radio.net/2008/03/episode-88-the-singularity-rese...
There is a difference between easy to set up and not having to set up anything. It's an improvement in operational UX.
It only changes behaviour if the risk of having to pay outweighs the benefit. And Google won't tell what they earned through that I guess.
I listened to an interview with one of the researchers. They had a component to verify "this binary will never allocate mem outside it's allowed area" and by statically verifying that they could enable high performance…
Sounds a bit like Microsoft's Singularity project, though I don't know if they even used a proof system to ensure a program couldn't do bad stuff(TM). After verification they just let the program run in Ring 0 (kernel…
> [int overflows, etc.] No runtime cost when Z3 can prove it. Otherwise, the compiler emits a safe runtime check as fallback. Super interesting approach. I see this eventually be integrated into future mainstream…
How I see this: Refactoring code to reduce the number of lines is _compression_, akin to RLE coding. Refactoring the code to lift conceptually coherent parts is _abstraction_. Less compression, more abstraction. Then…
Over the last year or so I arrived at a (sort of) MQTT semantic broker that facilitates an actor architecture. It supports federation (including transitive, so proxies "just work"(TM)), transparent outbound buffering…
Then the workers wouldn't spend 20 trillion and the economy as a whole would tank.
Why would one need a blackboard if one had a coordinating agent that would launch a sequence of builder agents that run either in parallel or in sequence depending on what the coordinating agent determined was safe?()…
Translation: As the number of bugs per line of code is a constant depending on language and project, we will experience a world full of bugs now that LLMs generate so much more code. On the other hand, LLMs are so quick…
I noticed back in the days that my FidoNet point client software ran a lot faster (actually processing a lot of messages, so IO and CPU heavy) in Windows than in DOS, which was unintuitive to me back then.
What was the minimum required density you derived?
Not if the models are way cheaper. For many tasks you can do with DeepSeek Flash, and for more gnarly problems you switch to GLM or Kimi. But sure, if you do everything in large models like Kimi K3 or GLM it gets…
Until the manufacturers enter an arms race and copy a page out of the mobile hardware vendors book. But perhaps they'll do it badly and we've got a few more years.
Apparently the task was to remove dead weight. Haven't read what they replaced "Enzyme" with. I did subagent based removal tasks a few times. These were the ones that required the least amount of input or thinking from…
Taste is something you use to build the guardrails of AI over time and many paper cuts endured. Once you have a big enough set of these (call them rules) recorded, they become your development substrate. You narrow the…
Shameless plug for my project, which is a bit different. I had the problem of crashing terminals due to OOM kills and wanted to switch to a more lightweight terminal. Also I had too many of them and couldn't find the…
OS is listed as FreeRTOS.
https://en.wikipedia.org/wiki/Lean_(proof_assistant) Not sure about "bugs" in that area, but there is a lot of work going on by mathematicians in formalizing and checking ever more complex proofs using proof assistants.…
And reading reports, endlessly.
If these proofs were output by the AI in a format readable by a proof verification system, the verification step of publishing vanishes. Then it's only valuable to check if the stated intention actually matches the…
Then it's a case of trying to uphold the narrative?
Tried it on Android and got "!!!!!!!!!!!!!" for answers.
Fair take if the user is the same person as the implementer. If I sell software to my customers that reduces their operational complexity by me investing some code, then I don't consider it a loss.
Can be that I remembered it incorrectly, as it was in 2008, but here it is: https://se-radio.net/2008/03/episode-88-the-singularity-rese...
There is a difference between easy to set up and not having to set up anything. It's an improvement in operational UX.
It only changes behaviour if the risk of having to pay outweighs the benefit. And Google won't tell what they earned through that I guess.
I listened to an interview with one of the researchers. They had a component to verify "this binary will never allocate mem outside it's allowed area" and by statically verifying that they could enable high performance…
Sounds a bit like Microsoft's Singularity project, though I don't know if they even used a proof system to ensure a program couldn't do bad stuff(TM). After verification they just let the program run in Ring 0 (kernel…
> [int overflows, etc.] No runtime cost when Z3 can prove it. Otherwise, the compiler emits a safe runtime check as fallback. Super interesting approach. I see this eventually be integrated into future mainstream…
How I see this: Refactoring code to reduce the number of lines is _compression_, akin to RLE coding. Refactoring the code to lift conceptually coherent parts is _abstraction_. Less compression, more abstraction. Then…
Over the last year or so I arrived at a (sort of) MQTT semantic broker that facilitates an actor architecture. It supports federation (including transitive, so proxies "just work"(TM)), transparent outbound buffering…
Then the workers wouldn't spend 20 trillion and the economy as a whole would tank.