choosing to believe that claude runs on a system that's heavily dependent on tens of trillions of 5.25" floppy disks and some poor tech that has to swap them when prompted.
i also was shit at learning home row when i was in school in the 90s and now type over 100wpm based on a technique I call "just typing a lot all the time for a few decades". when playing this game i had a hard time thinking of it as home row and more just replaying pathways in my brain, and it caused me to try to do things like pronounce "fjf jf" as "fijif jeff" and then type it from there.
what's the proce like for renting remote compute? i imagine it's a lot more than the overhead and just a bit more than bringing the chips in yourself. how satisfied is the market for renting compute from singapore? or to get to the question behind the question: can someone make more profit smuggling in chips than they can from renting compute from outside china?
>He made it clear in in extraordinarily forceful language that it is in the national interest for Deutsche Bank to emerge from the crisis and for us to become the leading player in Germany and Europe.
This sort of anticapitalist collusion between government and business, especially, finance, is just how things are done today isn't it?
the idea that a bubble means that when it pops AI will go away entirely is silly. the web was a bubble. we still have a web, and it's still useful, but a lot of people badly overinvested or invested in capabilities that never manifested and when the market corrected the price of those investments the bubble burst. same with real estate in 08-ish. A profitable idea drew more investment than it can support, prices went artificially high, then they came back down and we still buy and sell land.
I love terminal native games. I'm currently working on a real time transport sim that's native to the CLI and integrates with map services to calculate route time/distance.
I imagine equipment is regularly updated and upgraded without them knowing.
I'm extremely skeptical of this administration's competence, but this story is 100% finding anyone saying what you want to hear, however unreliable the narration might be.
> I imagine equipment is regularly updated and upgraded without them knowing.
Part of the reason for this outage was that these systems are not regularly upgraded. That's why they're dealing with decades old network hardware and lines and marrying them up to modern fiber. It's expensive to do the upgrades, but instead of doing regular rolling upgrades to their network components they generally do big-bang upgrades like the one happening now (that apparently caused the outage). In 20-30 years there will be another big-bang upgrade replacing the Juniper routers and fiber lines they've been installing now, and it'll probably cause similar problems.
This also happens a lot in DOD (not surprising, same contractors doing the work and probably the same companies in the program offices for FAA as DOD giving the same bad advice), it's incredibly frustrating because it's very obvious that there are better ways to do things, but the funding isn't there until it's urgent. That's why they're still using 1980s computers with floppy disks in their critical path for ATC as well. Now that those systems are nearly impossible to maintain, there is enough noise to push funding a proper replacement effort, about 20-30 years after it should have been done.
reply