I imagine a standard thickness is probably something like 9.5mm.

It's not all resin printing, he only used resin printing for one part in the video. Most was FDM printing. Though I did find it strange that he was using PLA. If it's parts facing repeated impact, you think he'd at least go with PETG, if not something like ABS/ASA or a nylon. I really don't get why so many people are so averse to non-PLA polymers. I guess PLA is more "eco-friendly"**, and yeah, there's a ton of PLA options out there, but that's mainly just because people are buying so much PLA.

** I'm actually thinking about switching some of my prototyping back to PLA, even though it costs more than PETG, because I can break down my PLA waste in sodium hydroxide at home. But I almost never make production parts in PLA unless that's the only practical option (for example, a given type of material only being available in PLA). Who wants parts that break if you look at them wrong, or melt in a hot car?

By "quite a large scale" you mean 1.3m long. He certainly could have made it longer than that. Honestly, I mainly think he just wanted an excuse to do more 3d printing and electronics work ;)

As per the principle in this article, that would make it worse if applied to the Reynolds regime in question. This is not about "general roughness", but specifically shaped roughness. In particular, a very sparse roughness on an otherwise smooth surface.

Sanding a hull is dealing with entirely different things. Sanding in general first off gets rid of microprotrusions and broader undulations. There is no question that this helps. The question to whether to polish to a matte or smooth surface is less obvious. Matte probably is better in general, as it helps make the surface more hydrophilic (there is also argued to be some potential to be making something like "riblets", although in practice you're unlikely to get the geometry right (true riblets are extremely thin walled).

This comment section is jam packed full of people mentioning the same misconceived and inapplicable thing, over and over. A thing that was actually discussed in the article, which they did not read.

How deep are the grooves?

A phrase to be heard either uttered on Slashdot in 2026, or at Woodstock ;)

But to be clear, the answer (assuming you're talking about sharkskin / riblets): a few dozen microns tall and a few dozen microns apart, with the individual riblets being very narrow, just a couple microns.

Hi, someone who regularly does CFD simulations in OpenFOAM here. It is a fundamental principle. I hope that helps.

This has nothing at all to do with how dimples help golf balls (just the opposite, actually)

This actually effectively "is" the paint. The only questions are about how durable it will be in flight conditions (two types are discussed, protrusions and dimples). Basically, paint with bumps vs. paint with nicks in it.

The dimples on golf balls are actually to create turbulent flow. TL/DR, a sphere isn't a very aerodynamic shape; its rear taper is too sharp, so flow detaches and there's a big low pressure wake in the back. High pressure in the front and low pressure in the rear = pressure differential, and a large area times the pressure differential = large drag force.

While it's best to not have flow separation, or at least delay it as long as possible, if you're going to have flow separation, you commonly want to generate vortices at the point of flow separation. That's why cars commonly abruptly truncate (kammback) where they'd become too steep in the rear rather than continuing to curve, and often have various vortex generating surfaces (lips, radial protrusions, etc) at the termination; it causes air to "pull down" and help fill in the wake. This is what the dimples on golf balls do.

Now, most of the dimples on a golf ball at any time are actually doing harm, or at least not helping. You really only want the dimples right around the point of flow separation. Unfortunately, golf balls don't have a specific flight orientation, so it's all or nothing - and "all" happens to be the better choice.

But as mentioned, this is entirely different than what is being talked about here, which is about the laminar-turbulent flow transition.

I think the only reading comprehension difficulty here is on your side. The impacts of "roughness" as a general term is a fundamental aspect of aerodynamic engineering. There has been evidence steadily emerging over time that this isn't exactly correct, that the distribution of roughness matters greatly, and the right distribution can even surpass a smooth surface. The confirmation in this paper helps close the chapter on this.

And honestly, their two approaches doesn't sound that difficult to manufacture at all. Certainly much easier than riblets. And the side effect of the first one - surface glass beads - would actually be beneficial for RAM. One of the principles for radar absorption is that you want a steady transition of the impedence (and by relation, dielectric constant) from the surface (which you want to be as much like air as possible) to the deeper layers. The outermost layer of RAM is commonly something like PTFE full of hollow glass beads. Under that you may have pure PTFE, and under a polymer with like 5% chopped carbon fibre fill, and so on. Well, here it turns out that having tiny glass beads on the surface can improve your drag coefficient as well.

AI video technology is still nowhere even remotely near just "click a button and take what it spits out". I don't know how to break this to anyone here, but you're not just going to go to some video generation site and turn out Woodnuts without extensive skill about AI video tools themselves and a wide range of traditional video production tools, and without spending weeks to months and significant financial expense on the project.

Even if / when this changes, video production is still always going to be limited by the human at hand. Most people's movie ideas, plotting, scripting, directing, etc frankly will be terrible. The slop in this case is the human, not the tool.

On the upside, AI lets anyone make a movie.
On the downside, AI lets anyone make a movie.

Including people who have terrible taste in plot, style, and everything else.

There's some genuinely good stuff out there - Gossip Goblin's work for example. But this is....

I'll just say, there's far better things that one could have spent half a million dollars on...

How are you defining "statistical inferences" as distinct from "logical inferences"? If you're defining fuzzy logic (e.g. not necessarily yes-or-no answers but allowing for ambiguity in conclusions), then we can agree conceptually, but your choice of wording is, I have to say, weird if so.

My understanding is that LLMs are built on a foundation of ANNs, and that indeed the backpropagation used to train ANNs is a statistical process;

Two responses. One, that's discussing individual-neuron scale processes rather than collective processes; and this was a discussion about inference, not training. Human neurons also learn by error minimization (Hebbian learning). But this does not describe the macroscopic processes that result from said minimization.

* During training, neurons develop into classifiers that detect superpositions of concepts that collectively follow the same activation process. Individual neurons weight their input space and subdivide it by a fuzzy hyperplane to achieve a classification result.

* In subsequent layers, said input space is formed from a weighted combination of the previous layer's classification; thus, the superpositions of questions being formed are more complex, as are the classification results.

* In a LLM, this iterates for dozens of layers, gaining complexity at each layer, to form each FFN

* The initial input space to a FFN is a latent (conceptual representation), as is the output; the FFNs, in result, function as classifier-generators; they detect combinations of concepts in the input space, and output the causally-resultant concepts into the output space

* FFNs alternate with attention layers dozens to hundreds of times in order to process the information, each layer building on the results of the previous one.

The word to describe that is not "statistics". It's "logic".

In a LLM, the first few layers focus on disambiguation. If there's a token for "bank", is this about a riverbank, a financial bank, banking a plane, etc? As the layers progress, it starts building up first simple circuits, and then progressively more complex circuits - you might get a circuit that detects "talking like MAGA", or "off-by-one programming errors", or whatnot. In the late layers, you have the general conclusions reached - for example, if it were "The capitol of the state that contains America's fourth-largest metro area is...", you've already had FFNs detect the concepts of fourth-largest metro area and encoded Dallas-Forth Worth, and then later taken that and encoded "Texas", and then finally encoding "Austin". And then in the final couple layers you converge back toward linguistic space.

Anthropic has done some great work on this with attribution graph probes and the like; you can detect what circuits are firing, and on what things those circuits fire, and ramp them up or down to see how it modifies the output. They very much work through long chains of logical inferences.