Interesting that for some people flow state is non-verbal. I personally have sort of a constant dialogue in my head (or sometimes muttered out loud under my breath) that I have to buffer or spool into various notes/diagrams/code. The process of prompting winds up being complementary to this—typing out that stream of consciousness into a prompt and editing it becomes a more effective form of reflection and ideation than my own process had been before. Sometimes I don’t even send the prompt—the act of structuring my thinking while writing it having made me rethink my approach altogether.
It’s a pretty large app (I think >1M lines of Python). I ran a self hosted version for several years and found it performant and pretty easy to keep running and updated.
That article is seemingly all about the perf of the complex frontend app with a custom renderer running in the browser, nothing to do with what’s happening on the server.
There are a variety of ways that democratic governments are structure that make this an inaccurate characterization of how things work.
The US, for example, apportions representatives and votes for President in a way that overweights less populated states, and there are various aspects of parliamentary systems that help avoid landing in a two-party system where a simple majority gets the say in everything—they force compromise and coalition building among disparate groups. Additionally, Constitutional systems will enumerate the rights of its citizens such that they cannot simply be taken away by a simple majority of any body.
Democratic countries are also basically never "pure" democracies where everyone votes on every decision as in your Plato's ship analogy—we elect people who audition for the role of running the ship, ostensibly those among the people who are best suited to the task.
I have not been following this whole thing closely, but this is where my mind went as soon as I heard there was some overlap in the popularity of this new un-sandboxed agent and people who are into crypto. It's like if everyone who is into buying physical gold started doing a Tiktok challenge to post pictures of their houses and leave their front doors unlocked.
Historically I've found sleepinginairports.net useful for this kind of intel—I was pleased to discover just now that it's still around! Nearly 20 years ago now it saved me from a night of awful sleep. My backpacking buddy and I had just arrived in Liverpool with an early Ryanair flight out the following morning, and no hostel reservations. We'd been traveling successfully without reservations for a bit, but it turned out there was some kind of event that weekend in Liverpool and there was absolutely nowhere to stay (or at least, nothing that cost less than, like, the budget for our entire trip).
We decided we'd just stay out late, then go to the airport and wait it out for our flight. After some effort trying to sleep on hard plastic benches in the airport Burger King (where Michael Jackson's Thriller was playing loudly on repeat, I do not know why), I pulled out my 12" PowerBook and found out via that site that the airport had a meditation room with dim lighting, soft carpet, and no Michael Jackson. Ahh.
Random events are the worst. I was driving through Bend, OR and planned to grab a hotel room but everything was completely booked. Ended up just driving up a random forest road (public land, legal to camp) and sleeping in my tent. Was walking around with my headlamp in the night and some cops came by and asked what I was up to since people are often doing drugs there. Great. They came by and woke me up during the night again and I asked them to please keep record that I'm not up to anything so I could sleep.
As a fan of text adventures who has played many over the years—Anchorhead is hard. It was kind of a white whale for me over many years until I finally beat it during the pandemic lockdown.
How does it compare in difficulty and scope to the original Adventure? I guess actually known as Colossal Cave Adventure? When I played it on my uncle's terminal in the 70s it was just called Adventure.
I stayed up all night and didn't get very far. I finally saw a solution online and I wasn't even close.
Historically retailers have employed buyers in charge of selecting products that would appeal to the store’s customers. A customer will likely have different expectations, and have an existing understanding of what sort of products they’ll find if they’re shopping at, say, Nordstrom vs Dollar Tree vs a guy on Canal Street in NYC.
Amazon sort of threw this out with the steady movement towards blending third party sellers in with products they sell directly. They made it less and less obvious and easy to filter based on seller over time, so now you have all sorts of junk from the digital equivalent of street vendors mixed with normal products, and it’s up to the shopper to figure it out. They tolerate tricks and fraudulent behavior from those sellers much more than they should.
Amazon could, if they wanted, make it easy to filter for products that have been selected by a buyer who has a relationship with the vendor, and are directly sold by Amazon themselves, but it’s seemingly more profitable to allow third parties to peddle garbage en masse.
For what it's worth, I evaluated Fly.io during a divorce from Heroku some time in mid 2022 (I think), found the platform was... way too rough around the edges at the time to want to migrate any real workloads. I kept it on my radar and shipped an MVP with it in 2024, found it was a lot more polished, and now have multiple production apps running there. I'm genuinely pumped about Sprites and have started building against the API—I did notice the weirdness with the docs, but you guys have been doing well on the "this thing that {was broken|I didn't like|was missing} now works the way I'd hoped it would" front.
Pure speculation, but I’d guess that an arrangement with Google comes with all sorts of ancillary support that will help things go smoothly: managed fine tuning/post-training, access to updated models as they become available, safety/content-related guarantees, reliability/availability terms so the whole thing doesn’t fall flat on launch day etc.
Probably repeatability and privacy guarantees around infrastructure and training too. Google already have very defined splits for their Gemma and in house models with engineers and researchers rarely communicating directly.
reply