pull down to refresh
Another example is that zig openclaw rewrite I posted the other day (#1445161). I did not look at the code at all nor did I have much sense of who the creator was or their credibility. I did have a rough idea of the concept (small, zig has a good reputation) and it didn't have glaring signs of slop.
I'm hopeful that there is a useful role posting such stuff can play: flagging projects for people who do want to look into it further. But your point really makes me wonder about the usefulness of amplification from someone like me who does not have the skills to do due diligence.
Perhaps I should just stick to bitcoin stuff where at least I have a rough idea of whether the idea makes sense (even if I haven't evaluated the code).
zig has a good reputation
Zig is a language. So you're basically saying: oh all is good because English has a good reputation? Idk about that one, haha.
Maybe we can do an uno reverse on the yolobois and build a bot that can be fed a repo and then it does at least baseline due diligence... don't know. I'll push some questions to Claude so that it can burn some more tokens. They need my consumption now that they aren't part of the war machine anymore.
While I am clearly in need of such a bot, I suspect there is no substitute for arming myself with a little better skill, or coming up with a new way to think about posting interesting things (=probably posting less github links I find).
There is a frantic-ness to things right now, FOMO or fear that I'm not producing enough value to be useful to people -- which leads to producing less value. I think you have pointed it out before, but this is a consequence of AI that I'm finding all over myself like some horrible rash.
this is a consequence of AI that I'm finding all over myself like some horrible rash.
That sounds really horrible! Hope you're otherwise ok!
But to stick with the Uno Reverse meme, perhaps that is an argument for tooling, not against. Despite what some stackers are claiming, the only way I see actual degrading of LLMs happening is when the runtime (GPUs) get messed with - either because of costs, interference (regulation, nationalization) or upstream catastrophes (power/water supply, weather) So that means that there is a capability now to (at whatever pace you desire) get some tooling in place to make your life easier.
Remember to take your time though; if the world is going to end it won't be because you didn't post enough content. There are other people plenty involved in tempting that right now. I'd just take a step back, because I know the fatigue is real. I spent 20 consecutive hours last week in a design->dispatch->review loop and I needed a full day to recover after that. It's ok. I got a shitton done. More than I would have hand-coded in a month. So sometimes... just scale down for a little while. Reflect on things. Productivity isn't worth much if you work yourself into a burnout in the process.
Right. Good points.
Super publishes a lot of prototype code, which is definitely not production code, but it is someone who we know does that and we can plan accordingly. So this is actually a perfect example: when you would post a super repo, it wouldn't even come to my mind to ever run that. It's just poc code for maybe an interesting idea that I (or Claude, or GPT, or Gemini) can then take as a starting point for building an actual production grade app or script, if I want it.
It's good to think about "what am I amplifying?". There is so much slop nowadays. None of it tested. So posting about it without having ran it basically puts your name next to something that might very well be completely awful. If you haven't tested it (or you don't have standards and just yolo'd it and have no idea that all your private thoughts are now sitting in a vault somewhere to be used against you in the near future) then what are you recommending?
For example, that picoclaw code that was linked here on SN and that I am looking into? I've changed about half the codebase thus far, and I still have tons of work queued up on it to make it even half acceptable as a poc that I can actually use. It's all slop. I won't publish the result because it won't help anyone that cannot re-code it themselves. In the end, it's just another example code that I really just should have converted into an implementation plan, to then write from scratch, bespoke. But I was hopeful (wrongly.)
If I see what stackers are doing right now, just jumping on things without understanding what they are doing (and not just stackers, the whole world has turned insane from FOMO) I'm getting really worried about the future. Years of very hard to get across privacy education are being wiped by stupid yolobots.