jerbear4328

The Windows app already existed, actually, as part of iCloud for Windows. It uses a Chrome extension for autofill. https://support.apple.com/guide/icloud-windows/set-up-icloud...

Well, the Passwords app is just the Passwords section in Settings moved out into its own app. It already exists on Windows, too, but maybe they are updating it to allow autofill without using a Chrome extension or add other features. It isn't the biggest change, just bringing attention to an existing feature that already competes with 1Password et al.

They automated away the part that had a human error (the internal tool with a field left blank), so that human error can't mess it up in the same way again. They should move that human labor to checking before tons of stuff gets deleted.

According to the wiki, it was not drawn on a computer, but created again in the same way as the first. I personally prefer the darker original, but most Windows users use light mode. It's cool that it has 25 people listed in the credits for what seems like a such a simple image.

> GMUNK was also involved in this version, and stated that it was created under the same methodology as the previous version.

https://windowswallpaper.miraheze.org/wiki/Hero#Later_versio...

Wouldn't that incentivize people to try to get companies to "ruin their lives"? If someone had a way to get a well paying income, then get a big company to do the bare minimum that counts as "ruining their lives", they live the rest of their lives comfortably without having to get a job.

I think that, in theory, compensating people for life-ruining is obviously the right thing to do. The issue is that it relies on people to be perfect and never try to abuse the system, and if we can assume that, then we can just get rid of law enforcement and become a communist utopia.

What is practical is making these companies far more liable than they currently are, though. It is unacceptable that they have no penalties for destroying someone's future.

Now make a cover for the microphone :)

Well, anti-stalking notifications for AirTags have existed on iPhones (and Androids with an app) since the beginning. This only adds built-in support to Androids as well. So, sorry to break it to you, but your system has always been set up to notify any thieves with an iPhone (or a free app installed). Anti-theft is an anti-goal for the AirTag, and, in my opinion, that is the correct choice, because vigilantism is really not a good idea. What would you do with the location of your stolen item? Report it to the police, they won't do anything, and you shouldn't go to that location, the risk isn't worth it.

And the audio clip (which doesn't work on the archive): https://static.nytimes.com/podcasts/2024/05/09/science/09tb-...

Nor is it how I work, I think that's normal enough. I do have an idea of what I'm going to say before I say it, I think that's closer to what they meant. I think and speak in increments of ideas, not words.

Unfortunately, no, OBS handles video only, no audio comes through the camera. It's really frustrating, I have wanted to do some more complicated audio setups on a Windows PC (specifically bridging two calls on different platforms together, I know the latency would be bad), but you have to use the VB CABLE proprietary driver, of which you can only have one for free, or try playing with Voicemeeter (same company with old, proprietary software). I have tried Synchronous Audio Router, but I couldn't get it to work after hours of pressing buttons.

On Mac, I can just buy Rogue Amoeba software, and on Linux, I can just run some commands or download a GUI to route audio around however I want.

Actually, it's not hashing either, it's just encoding. Anyone who knows the alphabet can easily decode them.

The news here is that gpt2-chatbot is confirmed to be an anonymized, unreleased model. See yesterday's discussion here: https://news.ycombinator.com/item?id=40199715 The https://chat.lmsys.org page now currently says this:

> *gpt2-chatbot is currently unavailable.* See our model evaluation policy [here].

That link takes you to https://lmsys.org/blog/2024-03-01-policy , which is this post's URL. The page was edited yesterday (when gpt2-chatbot was added) to include this section:

> Evaluating unreleased models: We collaborate with open-source and commercial model providers to bring their unreleased models to community for preview testing.

> Model providers can test their unreleased models anonymously, meaning the models' names will be anonymized. A model is considered unreleased if its weights are neither open, nor available via a public API or service. Evaluating an unreleased model consists of the following steps:

> 1. Add the model to Arena with an anonymous label. i.e., its identity will not be shown to users.

> 2. Keep it until we accumulate enough votes for its rating to stabilize or until the model provider withdraws it.

> 3. Once we accumulate enough votes, we will share the result privately with the model provider. These include the rating, as well as release samples of up to 20% of the votes. (See Sharing data with the model providers for further details).

> 4. Remove the model from Arena.

I'm interested to know how eScan, a security company, never noticed that they were using http to distribute executables to customer devices for so long.

What AI does this use? How can it be "end to end encrypted" if the server has to decrypt it to run an AI model?

Wow, that's really cool, could you share it? I would use that all the time.

Actually, Apple does this for iCloud! They use FoundationDB[1] to store billions of databases, one for each user (plus shared or global databases).

See: https://read.engineerscodex.com/p/how-apple-built-icloud-to-...

Discussed on HN at the time: https://news.ycombinator.com/item?id=39028672

[1]: https://github.com/apple/foundationdb https://en.wikipedia.org/wiki/FoundationDB

That whitepaper compares the counts of posts in a set of (arbitrarily chosen) hashtags to Instagram. All it proves is that the ratio of posts between these hashtags is different between TikTok and Instagram.

Their findings are unrelated to whether content is promoted or muted, they only look at post count. They don't even determine that it isn't Instagram that is biased (as they only compare the ratios), and these differences can pretty easily be attributed to different userbases.

They never explain how they chose hashtags that are compared, which could potentially be biased.

This isn't to say that TikTok isn't doing these things. This PDF just doesn't prove anything, and its claims are overstated.

See the discussion from this whitepaper was posted to HN at the time: https://news.ycombinator.com/item?id=38754653

I don't think the post was trying to prove that fact, just explaining a personal choice. I don't think every Rust project should have to include benchmarks to prove that Rust was the right choice.

Well, in Facebook's case, it's more than double (it appears to be 3 or 4 at most), and it is done smartly (the reverse caps try is only performed if you typed in your password with capslock turned on), and with a proper system to throttle attempts before locking an account, it seems reasonable to me. See https://security.stackexchange.com/a/214815 and its sources.

Facebook does do this, and it's perfectly fine: https://security.stackexchange.com/a/214815

It exchanges a few bits of entropy for a much smoother user experience, and it doesn't actually aid an attacker much (they could just try the permutations themselves)

No, you don't ignore the case, you swap it and try again. This could be implemented as a browser extension:

- User types pA$$WORD1 and clicks Sign In

- Frontend sends that to the server

- Server responds "incorrect"

- Frontend tries again with Pa$$word1

- Server responds "correct" and user is logged in

There is a link at the very bottom of the External Links section, but yeah, that is strange. Well, Wikipedia is the free encyclopedia anyone can edit.

If you want the more realistic introduction that introduced me to hexaflexagons, Vihart's YouTube videos are excellent: https://youtu.be/VIVIegSt81k

While this makes sense for teachers that are swamped with papers to grade, I really hope this isn't being used for grades without a human checking it. Now, using AI to give students personalized feedback immediately, that's really cool. I actually think Class Companion[1] does this really well.

[1]: https://classcompanion.com/

For people like me who didn't know what a "2-adic eval" is, it's a fancy way of saying that 52![1] can be divided by two 49 times and still give an integer.

If you factor the number 52! (https://wolframalpha.com/input?i=factor+52%21), you can see the term 2^49. If you divide 52! by two 49 times, you are still left with an integer.

[1]: 52 factorial, 52*51*50*...*2*1, the number of different orderings of the 52 cards in a standard deck

So can a human, yet we don't ban those. I don't think AI is going to get better at manipulating people than a sufficiently skilled human.

What might be scary is using AI for a mass influence operation, propaganda to convince people that, for example, using a weapon is necessary.

In 365 days, it will be February 28, 2025

Which is more important, the past or the future?

This is a good use of LLMs, if you check that the formatted code it creates is the same as the original code, it has been pretty good in my experience. It's especially handy when the code is in an unfamiliar language or library, and looks like gibberish to you, asking it to add comments makes things very clear.

Gemini won't even talk about Fox or Daily Wire anymore! NYT and CNN are fine though, hah. It's funny when these things are really obvious.

https://g.co/gemini/share/85464078bf89