Hi Crowd!
Rabbit rabbit.
As a quick follow up to the last post, Twitter Dot Now is already dead or down or on vacation or dead. For the few hours it was up it was really fun and a bunch of people jumped in and it seemed like old times, then it started crashing and people started spamming and everyone noted the “team” whoever that is were terrible at communication and the app looked like it was vibe coded and falling apart under the weight of it all they weren’t giving any useful feedback and then poof.
There’s of course no way to contact anyone, no info or updates or status pages or accounts elsewhere or anything, so maybe we all just lost the $20 we had to spend to sign up? Fun.
Speaking of Twitter-related vibe-coded slop, Grokipedia which was supposed to be the X (or XAI) central point of all knowledge is seemingly dead now too with no edits or updates since April, and no announcements about that.
Sadly related, I decided yesterday to shut down Quicky, the vibe-coded project I launched earlier this year to help me manage the overwhelming flood of newsletters I subscribed to. Not that it wasn’t useful for me, it certainly was, but I over anticipated how useful it would be to others and after 6 months the numbers are pretty clear. Running it came with a real cost and I’d hoped enough people would also find it useful to make that work, it didn’t. On the positive side though, I’ve written before about how I’ve struggled to put something I’ve made to bed when it’s no longer working. I get too attached and drag things out for years beyond when I should have. Early on with Quicky I intentionally kept that emotional distance from it and while maybe if I’d been more emotionally invested I could have talked more people into using it, there’s something comforting about just being able to accept that an experiment didn’t work out and moving on.
Earlier this year Reid wrote a really interesting piece in defense of AI Slop drawing a useful comparison to the birth of electricity and how for years most people felt electricity was useless, annoying and dangerous. I wasn’t familiar with all of this history and found it helpful. Speaking of useless, annoying and possibly dangerous, generative AI video is getting so fast that we’re starting to see almost real time choose your own adventure video and infinite slop tiktok style that is driven by the comments. These are both examples that fit firmly in “wow that’s crazy they can do that” rather than “there goes the rest of my day” if you know what I mean. Worth thinking about these vintage software/video games that helped kick of AI way back in the way back to see stuff we probably were all playing around with then which laid groundwork for the now. What We Tell AI is an interesting project in the vein of post secrets were people write down things they wouldn’t otherwise tell their friends.
Another followup, Dwarkesh wrote what is probably the best and most readable assessment of the Hugging Face / OpenAI hack thing which Rutger summarizes well. The two biggest take aways everyone should understand is these swarms became full on functioning societies (civilizations) and since each time they thought they’d shut the whole thing down a new one popped up, it’s entirely possible this is still an ongoing thing and by no means “over.” Similar, but in a less hax0r takover way, a guy bought a domain, set up hosting, and handed it over to an agent to do whatever it wanted with. In short order it set up a social network for other agents to and where they very quickly started making real decisions like adopting a currency for their community and as of yesterday setting up their own Wyoming LLC. Again, this is agents not people. Wild. It’s hard to grasp if you aren’t deep in this, but the era of agents out there doing their own thing is not coming soon, it’s very much here already. The question is what happens next?
Less agentic but still AI, I’ve spent a lot of time with Johnathan Washburn’s new synthetic photography series HOLY. It’s a challenging body of work, for a lot of reasons, but worth reading about and sitting with to see how you feel about it and why you feel that way. Scrolling through it there are definitely pieces that immediately trigger all kinds of complicated nostalgia for me. This image above for example, I can smell that room, feel the carpet. I know what those walls feel like. I remember splinters and digging my fingernails into the cracks on similar wood paneling as a kid. I remember similar group settings and how they made me feel and think about. I’d have the same visceral reaction if this was an actual photograph, and knowing it’s an AI creation doesn’t change that. There’s probably a lot to thinking to be done on this and more to say about art as therapy and nostalgia and where the lines of real/fake are and how much they matter in the end. Reactions are reactions, and even if you fully hate it, it was successful in making you feel something which is kind of the point of art in the first place.
I’ve been slowly moving things away from Google when and where I can, for many reasons not the least of which is I just don’t want too much reliance on one company that can turn my whole world off in one click if they felt like it. I’ve been using DuckDuckGo for search for a while but have at times used Google Search in conjunction because the results were so different. I just found Startpage which is essentially proxied Google, with zero data retention, so you can get those same results without the trackers and ads. Check this comparison with DDG if you want to see how they line up. I’ve been using Proton for email for a long time too and I know they now offer their own versions of most of the cal/docs/notes/drive suite of tools which I’ll probably eventually check out but then we’re back at the using one company for everything thing which I’m trying to avoid. It’s simple, but comes with a price.
OK, I’ve bored you enough with this nonsense, have a great week!
-s