nightpool

Did you even read the article? This is about offering companies a private 1:1 slack channel between your team and the support/sales team. Many companies offer this nowadays if you're on a large enough plan to be assigned an account manager and I prefer it >100x to email because I know a human is going to be in the loop.

Okay, but it's still a generalization. Most people don't care at all about the issues you mention

This is an interesting article to have up alongside the SLAP and FLOP vulnerabilities. I like capabilities as much as the next programmer, but my gut tells me that process boundaries are only going to get more important, not less, as chips get faster and untrusted code gets more widely understood. Or other sorts of hardward-enforceable memory boundaries

Commented where? I don't see any source links in the substack article.

No, just that what DeepSeek did is not as valuable as what the first movers did, because it did not advance the state of the art nearly as much. It's a new cheaper way to go from Base LM -> CoT "reasoning" LM. We already had CoT "reasoning" LMs, so while the new cheaper path to get to them is interesting, it's not necessarily groundbreaking either. Also, R1 only works with the "cold start" data that they distilled from o1, so it's not quite clear that it'll ever be able to exceed o1's capabilities. We already know it's much cheaper to distill new, smaller models from large already pretained and well-performing models—in fact, $5M sounds like a very expensive way to do so. So while these new techniques are probably going to have some impact, OpenAI is far from quaking in their boots

This would depend on your phone being able to permanently disable its radio, right? I don't know if I would trust my phone well enough for that, I would be worried even in airplane mode about it making some small beacon checks.

Are you using AI to respond to these comments? Yikes.

Small comment, but why does the text editor show an "overwrite" indictor instead of a cursor indicator when I try to change something in the playground? https://i.imgur.com/OAXbXQ2.png

I tried typing different things, and despite showing an overwrite indicator it seems to be in a normal insert mode. I have no idea which side of the highlight my characters will be inserted on though.

Of course it's rational behavior. Nico was the one claiming that they were just "following the law", that's what this subthread was about. If you agree that TikTok was making a political point by shutting down, then you agree with the person you're replying to.

Out of curiosity, I've always meant to ask, are you related to the famous Geoguesser content creator in any way? It's a pretty distinctive last name.

> Ultimately it comes down to force. The person with the pointiest sticks will likely be able to enforce their view about ownership over others.

This is a common but simplistic view that ignored e.g. concerns about popular legitimacy and support that often lead to the downfall of strongman regimes. Many people think they can enforce their views of ownership over others, but find that it's not quite that simple when they try to put it into practice. That's why I mentioned the social contract.

I'll be honest: I think it's pretty funny, and I think matt probably agrees. Also, as others mention, it's much easier/faster then removing the field and column and validation, yes.

No, they have auto-scaling as well. Not sure what GP was talking about: https://fly.io/docs/reference/autoscaling/

However, it's understandable that some people wouldn't necessarily want to pay for high amounts of max machines.

Apparently it doesn't auto-scale automatically either, you have to opt into it.

I think it's just acknowledging the reality that property is a social construct, one that's created by the social contract.

Surely you just use Docker for production, right?

Did you read the article? OP was very clear that OpenWRT fixed the issue in under 3 hours.

> The problem is that our system as currently set up does not incentivize reporters working a "beat" for years where they develop enough domain expertise on a field to report on it competently. If we had actual senior legal reporters, senior national security reporters, senior tech reporters, and so on, they would have enough knowledge of those fields to effectively write for laypeople.

How does this apply to Techdirt? Tim Cushing has been a staff writer at techdirt for 12 years and works heavily on a legal, civil liberties/policy beat with a tech perspective. By your criteria, he should be a perfect example of journalists who have enough knowledge to effectively write for laypeople.

Right, but what would be the security impact of that compared to just plain HTML? I guess it allows for some form of view counting or IP exfiltration, but other than that anything you can do with an external request you could do with an embedded data URI.

is the default-src necessary if you're using sandbox or is it redundant?

presumably "select count(*) from users" :)

In fact, apparently they now have 6.6M (just signed up for an account and saw what the number said)

Yes but it's just frustrating because in theory these APs are great—WiFi 7, 6GHZ mesh backhaul, 3x 4x4 MU-MIMO radios. But it feels like there was no QA put into the firmware and it struggles with reliability.

Our only local fiber ISP (Quantum Fiber, recently acquired by CenturyLink) just rolled new WiFi 7 routers to their entire subscriber base. However, I've noticed that these routers often just shut down when too many devices are connected to them, even though we have an extender. The router will get confused, and reboot, and then you'd have a few minutes of downtime while your computer tries to figure out what's going on

Tech on the phone says this is just how routers are supposed to work. He did clear my DHCP reservation table for me though, even though I was having problems with the APs and the DHCP / routing is handled by a separate box (afaik) and it was kind of a pain tracking down the 4-5 hardcoded IP addresses I had scattered around. He also says the new APs have no logging and no way for techs to access them. For a brand new system deployed in 2024! The router / fiber NID has extensive remote access and monitoring capabilities (although their mobile app has ~2-3 minute latencies for every page load). But no access to the APs! It boggles the mind. Just feels like box ticking all the way down.

This is a great comment ^ whenever you use two booleans like this, you're opening yourself up to "unrepresentable state" logic errors. Finding and noticing this in code can be tricky, but it's a great example of something that should be caught during code review.

Who would ever opt in from device settings if not given an incentive to do so? What would even be the point of providing such an "opt in" compared to just banning tracking entirely?

If you made tracking required in the device settings, you'd see the exact same pop-ups, they'd just give you a worse and worse user experience until you gave in and changed your device settings, or similar

Right, but the article doesn't explain why they cared whether it was keyboard or mouse activation. The linked WebKit bug goes into more detail, but it's still lacking an explanation of why alternative, more common and widely deployed strategies (like having a capturing keyup event that triggers earlier in the render loop) wouldn't be a better idea instead.

Also, if you really want to determine whether a MouseEvent is "real" or "synthetic", and you don't want to worry about when mouse events are triggered relative to keyboard events in the event loop (although it doesn't seem very hard to keep track of), it seems like you can use the current click count (i.e., event.detail). This works on both Chrome and Safari—it's 1 for mouse clicks, and 0 for keyboard "clicks", but the spec text is also a little contradictory and under-specified: the "click" event handler says that "the attribute value MUST be 1 when the user begins this action and increments by 1 for each click" (https://w3c.github.io/uievents/#event-type-click) but it also says "This MUST be a non-negative integer indicating the number of consecutive clicks of a pointing device button within a specific time" (https://w3c.github.io/uievents/#current-click-count), and the definition of "pointing device button" seems to exclude synthetic keyboard events (since those are handled separately)

Why are you filtering for screen coordinates in the first place? What if the user is using e.g. an alternative input device that doesn't have a screen? The `click` event should be enough indication that the user has tried to activate the menu. Why reinvent the wheel?

But they're already checking for event.name == 'click' in the revised code. So why would you want to filter out some legitimate click events?

Kinda weird that we don't get a graph for the final "solution"? I was looking forward to seeing how it compared to the other plots!

Right, and I'm saying "surely there must be a middle ground between "using all of Bluesky's infrastructure" and "having a 4.5tb copy of every post ever made on the network""

My understanding though is that having a PDS on its own is useless without an AppView to collect the data from the relay? Or am I misunderstanding the architecture here? https://docs.bsky.app/docs/advanced-guides/federation-archit...