You have a benchmark for output token reduction, but without comparing before/after performance on some standard LLM benchmark to see if the instructions hurt intelligence.

Telling the model to only do post-hoc reasoning is an interesting choice, and may not play well with all models.

I have seen reports of high-powered lasers being used to sweep an area, cutting through the fiber optic cables on drones that have passed by but not yet reached their targets.

But yeah, they do seem to always be future tech that just doesn't materialize as a reliable weapon system.

~100M*, but your point stands.

Not if you factor in the R&D

That's roughly the export price for them I've seen reported, which I would assume would cover some of the R&D.

Good lord, that is not how geopolitics works. You are supposing a normal market, not conditions of statecraft and network effect.

Why are you giving these companies the benefit of the doubt when they've already shown that they do this in the past?

I don't trust any of the cloud providers enough to store a single copy of important data. I always have backups go to two separate S3 style endpoints in different providers, and my infrastructure is setup with scripts of some sort.

I've run production setups on DO for 12 or 13 years now though. My last company, after it was acquired, was forced into Azure. Both stability and performance tanked, while spending a little over 2x compared to what it was in DO.

There are a whole lot of normal people using mesh networking Wi-Fi routers. Honestly, most of the least technical people that I know are all using mesh networks because their houses require it.

Certainly. But it's still a minority use case.

Perhaps someone else will (or did) write up a how-to for support mesh networking in your homebrew router.

Where do you live to consider mesh networking a minority use case? I live in a small city apartment so I don't need one, but everyone I know outside of the city needs at least two nodes to cover their houses.

I was looking at various stats and surveys, not going by my personal experience. But if you're asking about my personal experience, I haven't seen any consumer use of it at all, only enterprise and institutional use. That's part of why I wasn't going by by my own experience, because I know that the use isn't zero.

I don't live in a densely populated city.

Home mesh is mostly about having wireless backhaul, and you can certainly do that if you have (preferably) two radios, you just set up one radio as a client to your main AP.

Even if you aren't doing wireless backhaul you just rely on regular client behaviour to transition between APs, can enable 802.11r to improve this.

Enterprise "mesh" typically uses wired backhaul for performance and can help clients roam quicker with a controller (auth, not deciding to roam). Controller can also adjusts radio power so APs aren't talking over each other if they're too close.

Mesh isn't any magic, just regular wifi.

There are some difference in client wifi interfaces (STA) and access point wifi interfaces (APs, like you'd find on a good router). For example, some wifi interfaces don't have promiscuous mode, or can't scan while maintaining an active connection, etc.

It's like the difference between softmodems (aka winmodems) and full hardware modems. I know there are some projects that use Raspberry Pis as an AP, and it could do like 10 devices stock and 20 devices with firmware changes. Even a low-end router could handle more clients than that.

Oh heh, I thought they were asking about the X3D. My bad ><.

Or the vast satellite network we run. Pretty easy to see it's school children going in and out of the area.

To be fair, we don't really have the capacity to run satellite surveillance on each and every target we select to engage in a sneak attack.

I think sometimes people watch hollywood movies and get the impression that it represents a kind of cataloging of our military capabilities. A demonstration of what we can do to our enemies. With the underlying subtext being "don't mess with us."

I just want to gently suggest that not everything we see in movies is factual with respect to military or intelligence capabilities.

I'm an old timer. I got off the bus at Quantico in 1991. But even though I'm not in right now, I'd feel confident in betting that we don't have the capacity to surveil that many targets via satellite for, say, 1 week, prior to our attack.

(Of course, when I got off the bus at Quantico in '91 I also would have been just as confident in betting that the US would never engage in a first strike. So what do I know?)

That is true for an active war but I don't believe it is true if you have literally months and months to plan an attack. Unless of course there was no plan until just a few days before and you stupidly threw a ton of your advantage right into the trash.

So don’t sneak attack. Easy solution.

I remember being quite young and my parents going to the one of the local computer shops and getting a beige box Pentium 3 at 450mhz that we used for a while. The shop put Quake on there because they had kids, and I remember the first time I played it my mom instantly went and uninstalled.

A few years later in ~2004/5 I dug that same beige computer out of the closet, bought some extra RAM (I think it was 256mb total I could fit in it) and used that to host a private Lineage 2 server, which is how I got into databases / software development in the first place. With a whole bunch of tuning I could run ~50 people concurrently on that machine without terrible lag.

Eventually I had enough people who donated that I could upgrade to a newly released Athlon x2 stuffed into a rack mount case, which I sent to a colo.