RickS

Couldn't agree more. The vanilla openclaw install was this byzantine mess of MD files talking about souls and identities and such, it really put me off. Stripping back to a bare install of the underlying pi, it was delightfully minimal and easy to reason about. Excellent starting point for building an assistant agent without having to read or fight with a bunch of cruft on top.

Couple skills to integrate with an obsidian MD task tracker, small chat interface on the phone made public via tailscale, and bam, a reminder bot you can text from the grocery store.

Even if they didn't make a fair point (and I think they do), I would prefer their pedantry to your palpable contempt by a large margin. I mean "palpable" literally. My body is tenser and my heart more sad having experienced the voice you just used on that person.

Really rewarding keyboard UX on this. Typing felt great. I couldn't figure out how to shuffle or give up via the keyboard. Shuffle with escape would make sense.

Barring the retry thing, n=1 on all models? Am I misreading, or is this a joke?

Variance in quality on these things is so, so high.

US citizens to remain nonviolent at any cost, issue strongly worded internet comments, and find themselves a little less free every day.

Toggle the "prefer reduced motion" accessibility setting if on mac

I regret not signing up for Discord when they first introduced facial recognition and middle schoolers were trivially spoofing their ID checks with meme pics.

There's really something to be said for greedily signing up for most things and trying to get grandfathered before the zipcuffs tighten.

IRL, though, fuck this. Home depot added flock cams and broad facial recognition, grocery store installed turnstiles, haven't stepped foot in either since. I'm just dropping out of the IRL retail economy left and right.

If you watch the video (which is classic and the real content here), the lawyer cedes the second half to a career detective, who agrees with him in full.

What happened was the set of available moves and rewards has incrementally changed. What has not changed is that nerds are humans, and humans are terrible at resisting increases in status, power, wealth, etc. If the glorified nerds of the past had been offered the same roads to status and power, they would have taken them.

This reminds me of the way people think of the olden days when stuff was made of real wood and metal as "we had integrity! people built things to last!", projecting intentionality and generosity onto to the same machinery that built agent orange or rube goldberg machines for cigarette smoke to avoid liability for killing millions of people. We didn't build things out of metal because we had integrity, we did it because we didn't yet have advanced petroleum-derived plastics and shit. If they did, they'd have done that instead.

Reminds me of the difference between "peaceful" (capable of harm, electing not to be harmful) and "harmless" (incapable of harm even if you wanted to). I think it's a mistake to imagine the nerds of the past as peaceful. In terms of status, power acquisition, etc, they were harmless. Had you handed them the tools and understanding of today, they'd have acted no different, IMO.

It is hard. Much harder than regular software work. A few things need to line up: model iteration speed, task chunk size, and your own context window and comprehension abilities. Too slow and you lose interest. Too easy and it's just tedious riffing. Too large and you're burned out by reviewing giant complicated walls of text. Etc. The bites have to be the right size and speed for both you and the LLM.

In the few instances I've been able to achieve really joyful flow state, there are usually two simultaneous workstreams, plus or minus one. They're usually working towards a large goal that I roughly know how to judge, in digestible bites.

For example, sequentially modifying the UI in a series of operations towards an overarching goal, where it's easy to tell if a step worked, and what the next step should be, but where you're not sure exactly what you want, so there's some curiosity and discovery rather than just feeding the bot tiny instructions. I try to keep the two workstreams from overlapping. If both streams start fighting over a file they both dirty, things go south fast.

Adjusting your prefs/harness/etc for model terseness goes a LONG way. Context quality is absolutely everything. A good context "seed" can go back and forth for many turns cleanly and with focus, and even compact successfully once or twice. A bad seed will be annoying to work with from the jump, will thrash towards compaction faster, at which point it gets even worse. This is difficult to troubleshoot objectively, but I'll frequently restart conversations if I don't like the vibe of the first couple turns. It's made harder by constant model churn. Until opus 4.8, I ran opus/sonnet 4.5 high for a long time in large part for continuity of intuition, if that makes sense.

There are also many elements of human knowledge management that make a difference. I've found "append only" to be a magic word, generating markdown logs of changes, or learnings, etc. Whatever workflows create visibility and resumability so that you can return from a spell away and get up to speed effectively. Manually keeping your own dev log alongside the session sometimes helps, makes things sticky and ensures you understand what's happening.

But it's hard. Feels incredible when it goes well, but going well feels very nearly random, and whittling towards reproducibility can be very mentally draining, in terms of both energy and morale.

This just makes sense. A normal day contains dozens of experiences that could be bettered by cheap actions that I am awful at predicting or imagining. I had an argument with a partner at one point where I was baffled and basically at a loss, asked chatgpt, and it spit back a response that seemed... okay. I adapted it into my own voice, keeping only what was sincere, etc (not just dumping LLM slop at another human, which is fucked, more like using it to coarsely choose a vector/filter through a big cloud of things I actually believe). My partner's response was incredible. It completely diffused the situation and my they were pleasantly surprised. Without the LLM, I would have been entirely unable to conceive of and walk that happy path.

The problems we have with our psychology often involve deviation from the normal or desirable state, so a robot that spits out a cheap reversion to the mean can be really helpful.

My flavor of this is somewhat autism-coded, but it generalizes well. EG people who aren't used to negotiating, valuing themselves, etc. Obviously LLMs output hallucinated dogshit and occasionally dangerous nonsense.

But it must be admitted: a lot of our psychological hiccups can be solved by the thoughtless, typical advice.

38% putting them over professionals is nuts though. I would much rather have the real thing, but it's $200/hr and asleep at 3am.\

Worked great for me. Big recommend. "Cured" is mostly an unspecifiable state, and while certainly there's lots still wrong with me, I am healed far beyond my expectations at the outset, so increment your count by one.