frezik

joined 1 year ago
[โ€“] frezik@midwest.social 2 points 2 months ago (1 children)
[โ€“] frezik@midwest.social 10 points 2 months ago (3 children)

Five times more PVs at a 20% capacity factor means it evens out.

The vast majority of those PV installations were in the last 7 years or so. We also built more manufacturing capability during that time. Meaning PV installations are being rolled out on an exponential curve, not a linear one.

[โ€“] frezik@midwest.social 3 points 2 months ago* (last edited 2 months ago)

In a capitalist context, sure.

The idea of a socialist society is that there isn't a burning need to work beyond what's needed to keep life going. You can focus on art, or writing, or anything else creative. There's no particular need to legally protect what you create, because you're doing it for the pure enjoyment of creativity in the first place. Your livelihood isn't threatened by someone else copying it. If anything, you're delighted that someone else takes enjoyment from it.

And if someone wanted to feed your art to an AI model, that's fine, too. Who cares? That machine can't replace your personal creative drive. This is only a problem now because capitalism forces artists to make money off their art or do something else to make ends meet.

[โ€“] frezik@midwest.social 13 points 2 months ago

Here's my pet theory on the death of hard drives: it'll decline fast when SSDs hit twice the price per TiB, and we're only a few years away.

My thinking is that there's a lot of corporate deployments of four drive RAID 10. With SSDs, you can make that two drive RAID 1 with equivalent redundancy, but much, much faster.

Spinning platters are around $10 per TiB. SSDs around $60. (This does depend on the specific models of each.) SSDs would need to be cut in half about 1.5 times to reach that point.

[โ€“] frezik@midwest.social 4 points 2 months ago

That's always fun. Some people will swear it can't possibly work like that, but they have plenty of experience riding bikes. You wouldn't be able to turn properly at speed unless you're counter-steering, so they clearly have done this. The idea seems to be so incredibly intuitive that people don't even realize they're doing it, which is very interesting.

[โ€“] frezik@midwest.social 1 points 2 months ago* (last edited 2 months ago)

The idea behind clips is that you can pull up on the rising pedal instead of just following momentum from the other foot pushing down. It does work, but isn't really necessary for commuter biking.

I got a used bike that had a hybrid pedal with a clip on one side and flats on the other. While the clip (heavy) side usually landed down, it didn't always and it was weird to pedal with it. I just ride around on my bike, so I replaced them with cheap flat pedals and it's fine. I also converted it to an e-bike, and I don't need the extra pedal power.

[โ€“] frezik@midwest.social 5 points 2 months ago

It's pretty common to bring up gyroscopes for this when people know a little bit about physics. It's all over motorcycle forums, for instance.

As you say, it doesn't work. Experiments have been done where they attach a counter rotating wheel to cancel out the gyroscopic effect, and while it's a little wonky to ride, it works fine.

IIRC, we're not 100% sure how bikes work just yet. Every time somebody comes up with a model that seems to be good, someone finds a counterexample that throws it in the bin. Even your explanation of bike trail isn't all the way there; Razer-type scooters still work without trail on the front wheel.

[โ€“] frezik@midwest.social 4 points 2 months ago (1 children)

It works on its own. If you push your bike along with a good run and then let go, it'll stay upright until it slows down too much.

Learning to ride a bike is mostly about being confident enough to let the bike work itself out. It gets more stable as it goes faster, but it's natural to be afraid to go faster when it already feels unstable at low speed. Then there's a little bit to learn about countersteering, but most people figure that out without being told it's even a thing.

[โ€“] frezik@midwest.social 8 points 2 months ago* (last edited 2 months ago)

Pretty much. The US military can take on any other nation state (China is trying to change this, but it's not there yet). The initial fight against the organized militaries of both Afghanistan and Iraq didn't last long, and was as much of a one sided curb stomp as you'll ever see in history. It was the insurgency later on that was the problem.

[โ€“] frezik@midwest.social 6 points 2 months ago* (last edited 2 months ago) (1 children)

Neither is all that great in practice.

Gopher has many problems as a protocol. The original versions of HTTP had much the same problems, such as closing the connection at the end of a transfer rather than having a length header or a signal that the connection is actually done. HTTP went on to fix most of those problems, but Gopher never got the chance. Gopher+ started fixing it up, but it was a victim of bad timing. The Mosaic browser was released shortly after Gopher+ and everyone started switching over. To my knowledge, nobody has ever implemented Gopher+ on either a client or server. Not even after over 20 years of a "revival" movement.

Gemini intentionally limits things, such as not having inline images. This is supposed to be done to keep out methods that have been historically used to track users, but things don't work that way. I can just as easily send my logs to a data broker without using a pixel tracker if that's what I want to do.

In the end, you can just use HTTP with a static web page, zero cookies, and no JavaScript. That's what I ended up doing for my old blog (after offering a Gemini version for a while), including converting a bunch of YouTube <iframe> tags to linked screenshots so you don't even get YouTube cookies.

[โ€“] frezik@midwest.social 2 points 2 months ago

Attempting to replace people in the workplace without changing society so that people can live without work.

[โ€“] frezik@midwest.social 3 points 2 months ago (2 children)

Companies are expected to make money, not revolutionize the world

I'd like to believe that, but I don't think investors have caught on yet. That's where the day of reckoning will come.

AI is a field that's gone through boom and bust cycles before. The 1960s were a boom era for the field, and it largely came from DoD money via DARPA. This was awkward for a lot of the university pre and post grads in AI at the time, as they were often part of the anti-war movement. Then the anti-war movement starts to win and the public turns against the Vietnam war. This, in turn, causes that DARPA money to dry up, and it's not replaced with anything from elsewhere in the government. This leads to an AI winter.

Just to be clear, I like AI as a field of research. I don't at all like what capitalism is doing with it. But what did we get from that time of huge AI investment? Some things that can be traced directly back to it are optimizing compilers, virtual memory, Unix, and virtual environments. Computing today would look entirely different without it. We may have eventually invented those things otherwise, but it would have taken much, much longer.

view more: โ€น prev next โ€บ