Running ai locally is the same energy as playing a 3d AAA game for the same time

I wonder if they’re factoring in the energy usage to train the model. That’s what consumes most of the power.

I hope that goes well for them. It’s hard and extremely expensive, which is why there’s so few good search engines and half of them just use Bing’s API.

Do you have a source for that? I think it’s nowhere near 95% of sites given there’s several major providers that aren’t AWS or Cloudflare (eg Hetzner, Google Cloud, DigitalOcean, Wordpress.com, and a bunch more)

Definitely true. I’ll have to try it out. Is Ecosia better than DuckDuckGo or Kagi?

Google was best in the 2000s, but things were different back then. They were still a young company trying to improve the world, SEO spam wasn’t really a thing yet, there were far fewer websites, and most online discussions were archived and searchable (compared to today where there’s platforms like Discord that aren’t indexable in search engines at all).

I use it for document summarization and it works well. I use Paperless-ngx to manage documents, and have paperless-ai configured to instantly set the title and tags using Gemini as soon as a new document is added.

I chose Gemini over OpenAI since Google’s privacy policy is better. I’m using the paid version, and Google says data from paid users will never be used to train the model. Unfortunately I don’t have good enough hardware to run a local model.

It’s been common ever since magnet links were created, since you can post a magnet link anywhere (even in a plain text file) rather than having to upload a .torrent file somewhere like in the old days.

This is available in the UI too - there’s a tab labeled “Web”. Sometimes it’s hiding under “More”.

Adding it to the search provider URL is a good idea though.

Ecosia still uses American services though - they use Google, Bing, Yahoo and Wikipedia for search results.

I’m talking specifically about companies that post a loss every quarter, like Backblaze does. Backblaze’s net profit in Q4 2024 was -$14.38 million. Their net profit is going up year over year, but it’s still negative.

Even if the report is inaccurate, Backblaze has never been profitable, which isn’t great. Investors want to see a return on their investment. They’ll keep pushing Backblaze to become profitable, likely eventually resulting in enshittification of some sort. We’ll see if things have changed when they publish their Q1 2025 results on May 7.

One of the harsh realities in Silicon Valley is there’s a lot of companies that produce great products but end up failing after going public, either because they couldn’t find a good product market fit, or because they couldn’t figure out how to make their idea profitable.

I don’t have experience with rust on cars since I live in California and the main conditions where cars rust (high humidity, snow / salt on roads) aren’t a thing here.

I’ve got a 13 year old Mazda 3 that doesn’t have any issues though. No rust, and no major repairs needed so far. I’m getting rid of it soon (replaced it with a BMW iX) but it’s served me well for a long time!

In addition to Toyota Corollas and Honda Civics like others have mentioned, look for Mazdas as well. The Mazda 3 is a great car.

“I’m having fish tonight!”

A part of this was that they wanted people to drive far so they’d have to replace their tires more often.

A few basic steps can keep Arch just as stable as anything else.

“stable” in this case means “doesn’t change often”. Is that actually doable with Arch?

Debian testing is usually good enough. Packages have to be in unstable for ~10 days with no major bugs to migrate to testing. Of course, you can run unstable if you really want to live on the edge.

If you do run testing, you’ll want to install security updates from unstable, since testing isn’t officially supported by the security team. https://github.com/khimaros/debian-hybrid

I think they’re pretty different cases.

Amazon’s one was essentially a side project for them, likely fully funded in-house using their R&D (research and development) budget.

In Nate’s case, it was their entire product. They received funding from investors purely for the AI functionality that didn’t actually exist or work. They specifically claimed that it did work, which is how they got the money. They spent all the investor money and had essentially nothing to show for it.

They already factored in some amount of tariffs into the US price. It’s not really that it’s cheaper in Japan, but rather it’s more expensive in the USA. It’s also US$65 cheaper in Australia, for example, and even cheaper in the UK.

(keep in mind that advertised prices in Australia and the UK include tax, so you need to subtract the tax to compare with US prices)

The tariffs are just a lot higher than everyone expected. Nintendo were probably preparing for a 20% tariff, not a 54% one.

Maybe! I’m sure there’s loopholes of some sort.