Software engineer working on very high scale systems, and dad.

Born and raised 🇫🇷, now resident and naturalized citizen 🇺🇸.

🎹🎸🪕🥁🎮

  • 2 Posts
  • 6 Comments
Joined 1Y ago
cake
Cake day: Jun 17, 2023

help-circle
rss

Oh that’s interesting!

Yeah, that conversation is much, much older, pretty close to the very start of iCloud file storage. I’m guessing either things changed since and they used to be end-to-end encrypted, or more likely, what the friend was complaining about is his iCloud infrastructure team didn’t have access to the keys stored by another team, and reverse. So basically, Apple could technically decrypt those files, but they don’t by policy, enforced by org-chart-driven security.

Now excuse me while I go change a setting in my iCloud account… 😳


I once had a conversation under NDA (which has expired since) with an engineer at Apple who was working on iCloud infrastructure, and he was telling me that his team was a bit shocked to read that Dropbox was releasing apps for photos at the time “because they’ve noticed that most of the files users are uploading to Dropbox are photos”. He was like: how do they know that exactly? His team had no idea and couldn’t possibly find out if the encrypted files they were storing were photos, sounds, videos, texts, whatever. That’s what encryption is for, only the client side (the devices) is supposed to know what’s up.

Not having that information meant a direct loss of business insights and value for Apple, since Dropbox had it and leveraged it. But it turns out Apple doesn’t joke around about security/privacy.



Yup it’s been real. https://www.piquenewsmagazine.com/must-reads/bc-government-hit-tweet-limit-amid-wildfire-evacuations-7268169

The rate limits are because serving such a service at scale without the user noticing requires continuous innovation to get through scale bottlenecks; but with the engineering team greatly reduced, a lot of that work isn’t happening anymore. Typically, you’d get through those bottlenecks by coming up with some heuristics that make it seem like the service is doing a ton, when really it only needs to do little (like by sharding data, or by pre-caching a bunch of stuff). Without anybody to work on those heuristics to fake things, you gotta restrict with real restrictions.

Source: that’s what I do for a living. I’ve been working on some of the highest-scale services out there for over a decade.


No idea. I think they’re focused on being a newsletter, so my guess is probably not.


HackerNews being a bit of a time-suck, I’m subscribed to HN Digest, the daily newsletter of only the top links of Hacker News.


ChatGPT/GPT-4 mostly prove that humans are just continuous bullshitters
To be clear I’m not expert. But I know a bit. The way LLMs (like ChatGPT, GPT-4, etc) work, is that they continuously decide what the next best-sounding word might be, and they print it, over and over and over, until it makes sentences and paragraphs. And the way that next-word decision works under the hood, is with a deep neural net that was initially a theoretical tool designed to imitate the neural circuits that make up our biological nervous system and brain. The actual code for LLMs is rather small, it’s just about storing and managing representations of a neuron, and rearranging the connections between neurons as it learns more; just like the brain does. I was listening to the first part of this “This American Life” episode this morning that covers it really well: https://podcasts.apple.com/us/podcast/this-american-life/id201671138?i=1000618286089 In it, Microsoft AI experts also express excitement and confusion about how GPT-4 seems to actually reason about things, rather than just bullshitting the next word to make it look like it reasons, like it’s supposed to be designed to do. And so I was thinking: the reason why it works might be the other way around. It’s not that LLMs are smart enough to reason instead of bullshit, it’s that human’s reasoning actually works out of constantly bullshitting too, one word at a time. Imitate the human brain exactly, and I guess we shouldn’t be surprised that we land with a familiar-looking kind of intelligence - or lack thereof. Right?
fedilink

YSK: toilets and garbage disposals are some of the simplest devices to replace in a home
Both their functionings are complex, so people can get impressed. But for both of them, all the complexity is inside the device and there isn’t much to put together; and the way they hook up to your house is really simple. Why YSK: so you don’t live too long with a broken toilet or garbage disposal, thinking it will be too hard to replace. Those two are some of the simplest things to DIY.
fedilink