There’s a lot of debate over what the useful lifespan of the hardware is though. A number that seems very vibes based determines if these datacenters are a good investment or disastrous.
I specifically remember this debate coming up when the H100 was the only player on the table and AMD came out with a card that was almost as fast in at least benchmarks but like half the cost. I haven't seen a follow up with real world use though and as a home labber I know that in the last three weeks the support for AMD stuff at least has gotten impressively useful covering even cuda if you enjoy pain and suffering.
What I'm curious about are what about the other stuff out there such as the ARM and tensor chips.
Not exactly, it was a normal solar panel business started by Elon's cousins (SolarCity), but it wasn't going well, and in the end it was bought by Tesla for much bigger money than it was worth (let's say it was a bailout for Elon). Today Tesla solar panels are maybe 0.1%-1% of the business, they stopped giving any data on it years ago.
Seems like the real costs and numbers are very hidden right now. It’s all private companies and secret info how much anything costs and if anything is profitable.
That's like saying driving for Uber is profitable if you only take into consideration gas mileage but ignore car maintenance, payments, insurance, and all the other costs associated with owning a car.
Often when LLMs give me some command option or advice I haven’t seen before I try to independently verify it. And I’ve often been frustrated just how hard it is to find this info from the source documents.
Though a lot of the time this is more an inefficiency of the documentation and Google rather than something only LLMs could do.
As the rate of 'hallucinations' seems to have dropped dramatically (at least IME as regards non-existent flags and the like), I'm more concerned with usage. I often use grep.app/GH code search to look for usage examples as a sanity check when things look "off", for exactly the reason you described--there's often a total lack of good documentation on things like that, especially on "younger" tools/stuff.
No one would pay to have a 3D printer fixed. Or at least not pay enough for it to be a viable business. A brand new printer can be purchased for the equivalent of a couple hours labor, and that’s before replacement parts.
This is the problem (or arguably success) of modern appliances in general. They just aren’t generally worth repairing.
People grumble about planned obsolescence but the reality is that there are people who will repair them, if you are willing to pay. But when repairs cost a significant chunk of the price of a new appliance, most people opt to replace.
Covering something up is not bad faith. PR firms do it all the time (though plenty more do things in bad faith too). If what you're covering up is an explicitly user-hostile decision then maybe that's bad faith if what you're trying to do is trick people. But if you're just lying for brownie points then that's not always bad faith, just dumb.
I don't agree with your definition here. Good faith means trying to be correct but potentially not being by accident. Intentionally lying is bad faith and by definition trying to trick people; you know the truth is one thing, but you're saying something else to try to get them to believe it.
What I'm saying is that even lying is only bad faith depending on the intent of the lie. That doesn't mean others can't be upset regardless of the lie's intent, but I wouldn't say all lies are bad faith.
I'm saying this: I don't think a lie can be in good faith by definition. Trying to make someone believe something you know isn't the truth is fundamentally bad faith.
I thought you said Intentionally lying and bad faith is by definition trying to trick people. But you said Intentionally lying is bad faith and by definition trying to trick people.
I thought your point was intent. Most people would not say to hide Jews from Nazis was bad faith I think.
> Most people would not say to hide Jews from Nazis was bad faith I think.
That would be perfectly bad faith to the Nazis. There's no such factor as moral good or bad here; bad faith has more to do the intent you have towards each party. If the intent is to explicitly trick someone towards something you want or away from something you don't want, that is usually in bad faith. (There are some exceptions.) If the intent is just to explain something in a way others will understand, even if your explanation turns out to be (knowingly) inaccurate, that can sometimes be in good faith, though I wouldn't call it good practice.
Yeah, I'm pretty confused about what point they're trying to make. Given that a lie is intentionally saying something untrue, there are three possibilities:
1. A is lying to B, and they know that B doesn't know the truth. The intent is to make them believe the lie, which is intentionally misleading them and bad faith
2. A is lying to B, and they aren't sure if B knows it's a lie. The intent is to make them believe the lie, which is intentionally misleading them and bad faith
3. A is lying to B, and they know for sure B knows it's a lie. The intent is either to provoke an emotional reaction from either B or someone observing (which is bad faith), or performative for others who will see the lie and might fall into categories 1 or 2, which is bad faith
I don't understand how anyone could plausibly argue that lying to someone intentionally isn't bad faith. Maybe I'm the one falling for category three here
It mostly has to do with the potential effect of believing the lie and your reasons for telling it. If you know that believing the lie results in your advantage and/or someone else's disadvantage, then it's probably bad faith. If you don't know that however, and the intent is not necessarily to mislead, that's not always bad faith. You could argue that the very act of claiming a falsehood to someone is knowingly deceptive and therefore bad faith by definition, but I don't agree that's unconditional.
For example, if I lie to protect both myself and all other parties involved, that sometimes can be in good faith! It can be bad faith if I know that it hurts them and also know a less hurtful alternative, but if I really believe the less immediately hurtful alternative will lead to a worse overall outcome then I can still be acting in good faith. It's really a lot more nuanced than "deception bad". I have to deceive myself all the time to achieve good outcomes! now I wouldn't say my treatment of myself is good faith but I try sometimes.
They failed because traffic can’t be fixed by adding capacity. The inefficiency of cars will mean you can never build enough roads to keep ahead of consumption.
Traffic gets fixed by getting most people to use some other form of transport and leaving cars to the edge case uses.
Sure traffic can be fixed by adding capacity. Demand is ultimately finite. You see this in places like the midwest where there is overcapacity on the highway system and you can go a mile a minute across the region pretty much at all hours of the day.
Rental cars are expensive because they are covering the risk and increased wear and tear of rental drivers. As well as the downtime of when the car is rented to you but you aren’t driving it.
Self driving cars would make this massively cheaper and remove most of the reasons to own a car. It would make about as much sense as owning a train for most people.
how are you going to compare cars with no fixed location premium with housing? also utilization is totally different - cars sit ~95% of the time, housing can be nearly round the clock if you work at home. transportation is more like a utility/commodity whereas there are so many personal factors for where to live.
If you haven’t already, have a look at Immich. It’s a fantastic self hosted replacement for Google photos. They have pretty much perfectly replicated the UI.
Have you tried Ente.io and have any thoughts on comparison? I only use ente and have been happy with it but hear many good things about immich. Does it support E2EE?
I used Ente, switched to immich just recently. Overall immich gives more of a quality feeling, has much more community support as well and a clear roadmap. I also think it will eventually receive nice, native apps for all platforms with the support of sizeable community. E2EE in Ente was nice but that's pretty much its only advantage.
Immich is self hosted only so it doesn’t really need e2ee since you can just encrypt the disk of the server. It also runs a load of on server machine learning stuff for automatic people tagging and search.
Ente is selfhosted (also has a hosted version) but encourages family use so I think that's why they do E2EE. It also does all the ML on the client side for people tagging and search.
Mozilla backed it with a grant but that was a few years ago.
reply