addaon

I assume this is what's up with the NordVPN ads flooding YouTube and stuff. No idea who the buyer is, but if I was in the market for a VPN I'd presumedly do a web search, skip NordVPN as having poisoned the well of good will, and choose whoever's next.

putting AI queries into ever search response

There's no way these are using a significant amount of compute. I'm not 100% convinced they're actually LLM-generated rather than an old-school Markov model. Both the relevance and accuracy numbers of the responses flirt with 0%. It's possible they just have a few million stashed responses and choose one at random, from what I can tell as a user.

The toughest part was getting the Go-written runtime to be able to run arbitrary code in multiple languages.

Why not just supply a Wasm VM? Are there interesting languages that don't compile to Wasm?

Civ 7 still isn't good, but Test of Time made it somewhat better. It took a long time for Civ 6 to get good after release, but I'd argue they got there. Still frustrating, though.

Booting on what? Are BIOS calls (if allowed) considered a library? The overhead of managing a framebuffer etc without an OS (which would be a required component not included on the disk) feels like it would dominate.

My recollection is that Civilization (1) came on 5 (? 7?) floppy disks for Mac, but only the first disk was needed for the full game in black and white; the additional disks were needed for color and some add-on assets like improving your palace.

"Build a game that stands the test of time as well as Civilization" is a pretty high bar. 1.44 MB is huge.

That's a real risk, yes. There's definitely a difference between cases where the code is the artifact being built, and cases where the code is being built to generate an artifact that is independently valuable and assessable. I'm happily in the latter case -- but I definitely share your fear of those who think current LLMs are valuable in the former case.

It's hard to say if this code is structurally better or worse than before, but it's certainly voluminous and as far as leadership could ever tell, with their flawed metrics, that is all that matters.

Hah. I've been working with ChatGPT-5.6 Sol and a lot of guidance to port a particular geometry from 2800 lines of Python-scripted Rhino3D (basically a custom parametric CAD kernel in there, plus use of it) to Python-scripted OCCT/FreeCAD (an existing CAD kernel) and it's up to 36,000 lines and only about half way there. And I've been setting goals and targets for duplication minimization, code size reduction, etc. The results are fine for my purposes, but if I put a positive value on "voluminous" I'd be super impressed; with my actual negative value assignment on lines of code, this is more "absolute crap but still useful to me."

I think "absolute crap but still useful to me" is a pretty high value and not worth neglecting, but I'm definitely scared by leadership who sees the toilet overflowng and assumes that means that value has been maximized.

That said, I really wish there was a "I know what I'm doing" checkbox to avoid a lot of that BS...

The "I know what I'm doing" approach is the absence of the checkbox. Each of these decisions matters, and gives you insight into what the programs you're installing are trying to do today, independent of what they did when you last installed them (and maybe audited them) five years ago. Neither default behavior of "disallow all" (which would prevent correct operation of many programs) or "allow all" (which would likely violate your trust assumptions) is safe -- thought and human decision-making based on your own risk model and your own intended use of the installed software is needed.

more evaporation over oceans, which makes the air denser

More humid air is less dense than less humid air at the same temperature and otherwise same composition. H2O has a molar mass of 18, vs ~29 for dry air.

there is no "software" that a lot of people want, yet nobody managed to create yet because they failed too due to it was being hard to implement (excluding AGI/ASI which is not really software)

What!? I can think of about a billion examples... but for one, I'm still waiting for a good enough CFD/FEM coupled system to model paraglider dynamics across collapse/recovery. And I expect to be waiting quite a while.

At transonic speeds, wave drag dominates both induced and parasitic drag. The hump reducing drag is about shape (area rule).

how will Tidal consistently determine AI generated music?

Is this their responsibility? Just restrict payment to the registered copyright holder or their delegate, require registration of copyright for music to be payment-eligible, and escalate the problem to a federal crime with (presumedly) federal enforcement, no? Sure, some people will commit federal crimes to get a payout, but it's gotta reduce the problem massively.

Since when are we doing 32-layer planar transistor logic on a single chip? Even ignore the use of FETs for eDRAM… I didn’t realize we had decent logic density possible on BEOL.

Top floor existed at all because it was Boeing's entry for a heavy cargo plane competition

Yes, but it turns out the hump is great for area ruling (aerodynamic drag reduction at transonic speeds), as observed by the 747-300's extended hump giving lower drag (but higher weight, of course) than the short-hump versions.

Woulda given the AI credit if it had gone with "The merchants of Assur called it Nergal."

Meanwell is the standard answer for this sort of thing; something in the SD-500 family or sized/optioned as you need. You'll have to do the connectors yourself; you may be able to find junction-post-to-barrel-plug leads of the right size and length off the shelf, but I'd be surprised, and soldering them would take less time than shopping for them.

Why use the original power bricks, with the space claim and awful routing, instead of just going to a single dc/dc... either directly if no individual power control is needed, or to a relay block or switch block if automated / manual individual control is needed?

Me too!

it's unlikely that you'll get people that are in overlapping networks

Networks all overlap if you search deep enough, ask Kevin Bacon. For good-enough talent and starting with random nodes, yeah, you'll probably not overlap. But there's a few dozen people of top talent for any particular role you're looking for, so by the time you iterate to the top of the tree, you really do get a lot of repeats.

Yes, making people an offer they can't refuse isn't scalable. It generally requires both deep pockets and the time investment to understand what would motivate them to leave a good job for one that they think can be great.

I am aware of several startups that started this way, and have been involved in some. The quality of the results depends on who you seed the graph with, of course; but I’ve seen it work well.

As requested by the original poster, it doesn’t scale.

Ask the best people you’ve ever worked with who the best people they’ve worked with are. Recurse. When names start repeating through different graph paths, make those people an offer they can’t refuse. Once they join, ask them to do the same, and give them the budget and role to make it happen.

costs are only likely to decrease over time

> There was no argument that the price will forever keep going down

It's very hard for me to read the first quote as anything other than a continuing decrease in expected value of cost as a function of time. This directly contradicts the second quote.

In order of safest to least safe: commercial jets, automobiles, general aviation aircraft, motorcycles, messing around with Jim.

Everyone is pointing out examples around image reproduction, which are valid and interesting… but the case that comes up in nature is violet (beyond blue in spectrum) vs purple (mix of red and blue) pigments.

But that's my point. If you're basing your society shape around adopting a technology based on it continually decreasing in price, but you only get a few decades of that behavior before saturation and then you're at the mercy of the consolidated winners... generally adoptions like this aren't reversible at the societal level. You're locking in a long-term structural change based on a short-term pricing trend.

it’s costs are only likely to decrease over time

What is this based on? We're well past a 50-ish year deflationary period in the cost of major appliances (refrigerators, washing machines, etc). We're pretty clearly at or near the end of the deflationary era for computers and computation. Automotive... speaks for itself. We're still there for televisions, surprisingly; but it looks like these technologies tend to have a handful of decades of rapid cost decrease, followed by a never-ending cost increase over time as the manufacturers consolidate and claim an ever-increasing margin.

Okay, that video is great.

Product questions that I couldn't find an answer to. From https://www.crowdsupply.com/modos-tech/modos-flow, I see "On the go, you can power Flow at up to 40 Hz with a single USB Type-C cable. At a desk, you can connect additional power and take advantage of its full 60 Hz refresh rate."

1) This surprises me a bit... is USB-PD incompatible with DisplayPort alt mode, or is this just based on an observation that display port devices tend to give limited power output?

2) Is every DisplayPort alt mode host able to give enough power to run at 40 Hz? In particular, can this be driven on the go directly from an iPhone?

3) Is that second USB port usable as a data port hubbed to the device when powering over the DisplayPort port?

4) I know it's possible to provide power from the display back to the host device when using DisplayPort alt mode -- when powering the display from the second USB-C port, is the connected device also powered?

The two use cases that would be super interesting to me is plugging this in to my iPhone or similar on-the-go, and plugging a USB-C keyboard into the second port on it for quick e-mails at the coffee shop and similar; and plugging this in to an iPhone, plugging my power bank into the monitor and keeping the monitor in high-power mode and the iPhone charging while working with a Bluetooth keyboard.

Obviously I don't expect it to handle these use cases out of the box, but... open source! This is really a question about what the hardware design is capable of, not the current software/firmware/FPGA capabilities.

They’re not dead until they’re warm and dead.