But beyond that, shutting down a subreddit for multiple days is such a drastic action, it should require more than just a simple majority in a quick online poll.

For instance, to amend the Constitution, you typically need a supermajority (2/3 or 3/4 of different parts of the government). To convict someone guilty in a trial, you typically need evidence beyond reasonable doubt and a unanimous jury verdict.

The burden of proof that users want to shut down the subreddit should be overwhelming.

This isn't an accurate extrapolation. In fact it's extrapolating in the opposite direction.

The moderators are taking a moral stand against an API policy. But the the actual users of the subreddits never consented to this moral stand. By removing these moderators that are working against the users, reddit is giving the subreddits back to the users.

If a competitor magically got the support of all the moderators, that wouldn't get them very far.

The process for selecting moderators is way less meritocratic or democratic than this. They merely got there first, finders keepers. The analogy for landed gentry is accurate.

- Product wireframes/mockups

- Memes/social media posts

- Simple vector creation/editing

And that list is only expanding.

They already do it for iMessage, and it makes it easier to turn down subpoenas if they can credibly claim that they can't even access the data themselves.

Likewise, offering an explicit choice also seems plausible. The full E2EE flow might have UX downsides (for example, the user might need to write down a recovery seed phrase on paper), so they might not force all users into that flow.

I really enjoyed it. Once I applied the powder, everything in my fridge tasted like candy - fruits, yogurt, etc.

Now whenever I'm hungry for a snack, I quickly apply some of the powder, and have a delicious, healthy snack.

For me, the bottleneck to picking up vim has not been merely knowing the hotkeys, but more internalizing the commands and building muscle memory.

The interactive game format of this website could possibly solve that, so it appeals to me.

As in, rather than learning in depth all the low level parts then finally putting it together at the end, start with a surface high-level understanding of a working prototype then expand into the details of how everything works inside.

In the case of ML, this could mean starting with a 5 line SciKit-learn prototype of a random forest model, seeing some working predictions, then expanding knowledge from there - what data is going in and what is coming out? What’s a classifier? What’s a decision tree? Etc

That would be like calling your iPhone fragile if it breaks after throwing out the window of a skyscraper.

I interpret end-to-end encrypted to mean only the participants have access to the content, the intermediary couldn't spy on my calls even if they wanted to. This appears to NOT be the case with Zoom.

For instance, if a team loses a sporting event by 1 point, and their Gatorade was poisoned, it would be understandable to call out the Gatorade poisoning. Of course, many other things could have swung the outcome as well (random variance, wind, poor effort, etc).

Keeping most of the 200 employees on payroll may have been a condition of the $3m sale price.

https://www.fast.ai/2019/03/06/fastai-swift/