That's the fun part! You spend all day hardening it... run it in docker in a vm on a separate machine. And then you hook it up to your gmail and give it unrestricted internet access :)
Cool idea, but this took about 15 seconds to load for me and then lagged very hard, especially while zooming. So I wasn't able to use it very much before getting frustrated enough to exit the page.
That's it! There are no other source files. (Of course, we outsource the agent, but I'm told you can get an almost perfect result there too with 50 lines of bash... watch this space! (It's true, Claude Opus does better in several coding and computer use benchmarks when you remove the harness.))
Fair enough. Is that all it takes? The heartbeat is only a few more lines of code. Cron made the code a few hundred lines instead of a few dozen so I didn't like that.
(Also, I think heartbeat.md can emulate Cron? Using an LLM to expensively and inefficiently emulate Cron sounds a lot more in line with the Claw philosophy, doesn't it? ;)
The neat part is that it can modify/upgrade/restart itself. So if you are missing any feature, you just complain and it adds it to itself. (And it does that more reliably than OC in my experience, because it's small enough to actually understand itself.)
I have the same thing on YouTube. I usually use adblock but I used youtube without adblock recently and was startled by the ads. It's either "AI girlfriend", or video games, or video games about AI girlfriends. (I don't play video games, and I'm not interested in AI girlfriend. At least Meta shows me ads about stuff I actually find interesting!)
But my experience is constantly interrupted by images of scantily clad AI generated women. I'm no prude but it seems more than a little inappropriate to me.
Oh yeah and the other 10% of the ads are about exploding children.
So there's a strange situation with the incentives here.
The whole point of the AI posts with the AI bots in the comments section is they're waiting for the one clueless actual human to show up so they can scam them.
But here, the obviousness of the scam is a feature. Just like, obvious scams in the spam are actually a feature because they select for the most gullible people to scam.
So this is funny because, the shittiness of the AI images is actually a feature in the same way. The continued improvement of image generation models will actually worsen the situation from the scammers point of view, because they won't effectively filter for dummies anymore.
P2p is generally bad for this usecase. P2P generally only works for keeping popular content around (content gets dropped when the last peer that cares disconnects). If the content was popular it wouldnt need to be archived in the first place.
IMO there is actually a very low hanging fruit here, even without P2P or DHTs we could have an URI scheme that consists of a domain and document hash. It is then up to the user to add alternate mirrors for domains. Aside from privacy, it doesn't really matter who answers these requests since the documents are self-signing.
I'd say taste-as-subjective-something is largely irrelevant. If something "looks good" but hurts to use, that's not much help. If it looks like ass, but is a delight to use, that's not good either (because most people won't reach the point of actually experiencing it). So you need "looks good" and you need "actually delightful to use". Taste seems to be orthogonal to both of those. Or perhaps (two kinds of?) taste is involved in each one.
At which point we define taste as two unrelated things: skill in aesthetics, and skill in ux.
I've seen apps that looked amazing (Taste #1, aesthetics) but made me go, "Okay, did they actually try using this thing?" (Taste #2, usability). I think these tastes are completely orthogonal, from personal experience. I think the vast majority of designers suffer from Total Usability Taste Blindness.
(And, though it feels a bit mean to point out, the vast majority of FOSS suffers from a total absence of both. The winning projects only win because they have no competition, they're the only free option available.)
So, OpenClaw is literally a meme. It existed for months and didn't get much attention until it became a meme. And every day, you see people fell for the meme wholesale: "I bought a Mac Mini for OpenClaw, now how do I run GPT-5 on it locally?"
The technical characteristics appear to be entirely irrelevant. (I'm not sure if taste even enters into the picture.. it appears to be a third category!)
reply