jofer

More likely, the date attribution in the imagery is incorrect.

As someone who works in that exact field (literally - I produce seamless mosaicked global maps and work for a satellite company), I can assure you that we don't and can't do this (generate "fake" imagery). Depending on country the exact satellite is licensed through, some areas may be lower-res (e.g. sats licensed via Canada can't provide imagery in active conflict zones above a certain resolution). The US govt can in principle demand we stop imaging for US-licensed satellites (though they never have so far as I know). A lot of regulatory details can vary based on the country 1) your satellites are licensed in, 2) your company is based in, and 3) where you're selling data.

However, none of the imagery is "fake". Our imagery sometimes feeds into google maps, and I don't know google's exact processing chain, so I can't rule out them doing something like that. However, I'd be absolutely shocked if they were for a lot of different reasons.

It's _way_ more likely that the tile data is incorrectly indicating 2025. E.g. they're using 3rd party data that doesn't have detailed scene provenance information in that area and are just showing "2025" in the absence of other information.

More interesting is the things China and some other countries do around datums. If you process things correctly, your satellite data won't align with their official street/infrastructure maps. Instead it will be randomly and smoothly shifted in different directions across the country. That's to make on-the-ground targeting based on official maps much more difficult. E.g. you can't reliably take one of the official maps and go "point the artillery at an azimuth of 321.5 degrees and target a location 4567 meters away". However, it makes things really tough when you're trying to provide a correctly processed "backdrop" mosaic to Chinese customers. (IIRC, this problem has gone away due to the ubiquity of OSM data or regulations changed in recent years. Still, China in particular has a lot of interesting regulations around map accuracy.)

In addition to disease, a key issue in many climates is toilet paper. Your average deer isn't leaving around white paper that takes a decade (in dry climates) to go away. That's a non issue in wet areas, but a large one in deserts and more arid regions.

If it helps, here's some recent coverage:

https://www.theguardian.com/world/2025/aug/26/indonesia-prot...

As much as X11 is an overly complex and dated protocol, this article hits its mark well. My current desktop is actually running Wayland, but I still need X11 for a variety of reasons.

I see the development advantages of Wayland, but not the practical advantage as a user. And even as a developer, X11 is stable and well known (albeit definitely weird in places).

At the end of the day, things worked perfectly on X11 and my audio and video and various apps still glitch a lot on Wayland even after all these years. Most of that is not exactly Wayland's fault, but it highlights the advantage of X11. It's the devil you know (and everyone has worked out a lot of edge cases for).

To be fair, it's the higher level folks who are too far removed from things to have any actual authority. I've never heard a direct single-team engineering manager something like that. But yeah, CEOs say crazy crap. And we're definitely there, though to be fair, his exact quote was "I insist everyone try to have AI generate your code first before you try making any direct changes". It's not _quite_ as bad as what I described. But then the middle management buys in and says similar things. And we now have a company level OKR around having 80% of software engineers relying on AI tooling. It's a silly thing to dictate.

I appreciate this writeup. I live in the terminal and work primarily in vim, so I always appreciate folks talking about tooling from that perspective. Little of the article is that, but it's still interesting to see the workflow outlined here, and it gives me a few ideas to try more of.

However, I disagree that LLMs are anywhere near as good as what's described here for most things I've worked with.

So far, I'm pretty impressed with Cursor as a toy. It's not a usable tool for me, though. I haven't used Claude a ton, though I've seen co-workers use it quite a bit. Maybe I'm just not embracing the full "vibe coding" thing enough and not allowing AI agents to fully run wild.

I will concede that Claude and Cursor have gotten quite good at frontend web development generation. I don't doubt that there are a lot of tasks where they make sense.

However, I still have yet to see a _single_ example of any of these tools working for my domain. Every single case, even when the folks who are trumpeting the tools internally run the prompting/etc, results in catastrophic failure.

The ones people trumpet internally are cases where folks can't be bothered to learn the libraries they're working with.

The real issue is that people who aren't deeply familiar with the domain don't notice the problems with the changes LLMs make. They _seem_ reasonable. Essentially by definition.

Despite this, we are being nearly forced to use AI tooling on critical production scientific computing code. I have been told I should never be editing code directly and been told I must use AI tooling by various higher level execs and managers. Doing so is 10x to 100x slower than making changes directly. I don't have boilerplate. I do care about knowing what things do because I need to communicate that to customers and predict how changes to parameters will affect output.

I keep hearing things described as an "overactive intern", but I've never seen an intern this bad, and I've seen a _lot_ of interns. Interns don't make 1000 line changes that wreck core parts of the codebase despite being told to leave that part alone. Interns are willing to validate the underlying mathematical approximations to the physics and are capable of accurately reasoning about how different approximations will affect the output. Interns understand what the result of the pipeline will be used for and can communicate that in simple terms or more complex terms to customers. (You'd think this is what LLMs would be good at, but holy crap do they hallucinate when working with scientific terminology and jargon.)

Interns have PhDs (or in some cases, are still in grad school, but close to completion). They just don't have much software engineering experience yet. Maybe that's the ideal customer base for some of these LLM/AI code generation strategies, but those tools seem especially bad in the scientific computing domain.

My bottleneck isn't how fast I can type. My bottleneck is explaining to a customer how our data processing will affect their analysis.

(To our CEO) - Stop forcing us to use the wrong tools for our jobs.

(To the rest of the world) - Maybe I'm wrong and just being a luddite, but I haven't seem results that live up to the hype yet, especially within the scientific computing world.

It's always been active. The Sagaing fault is a plate boundary. You're seeing the "side" of the Indian subcontinent slamming northward into the Eurasian plate.

These same things apply to satellite images of the Earth as well. Even when you have optical bands that roughly correspond to human eye sensitivity, they're a quite different response pattern. You're also often not working with those wavelength bands in the visualizations you make.

Scientific sensors want as "square" a spectral response as possible. That's quite different than human eye response. Getting a realistic RGB visualization from a sensor is very much an artform.

The moon itself already is one. Moonshots are widely used in calibration, at least for earth observation satellites. The brightness of the full moon at each wavelength at each day of the year is predictable and well-known, so it makes a good target to check your payload against.

In case you weren't already aware, that last bit basically describes most optical scientific imaging (e.g. satellite imaging or spectroscopy in general).

Similar tests have been standard for over 20 years. When I worked at McDonald's (late 90's), they didn't do the personality test, but when I applied across the street at Arby's a few years later, they did.

The one that I just got annoyed with and decided it wasn't worth switching from McD's to Arby's was "would you rather read a book or talk to a person?". I mean, I get it, they want people-focused-people, but being introverted and/or just liking books doesn't mean you can't give excellent customer service.

Sure, it's easy to guess what want most of the time, but the fact that personality tests are as widespread as they are in employment is maddening.

Many years later I worked at Chevron (upstream as an exploration geologist -- not a gas station). While they didn't do it as part of the application process, you were required to take a personality/communication style test when you started (ecolors). That's all well and good (it _is_ very useful to understand personalities for communication styles), but in a lot of roles you literally had to wear the colors on your badge. If you wanted to go into management, you essentially had to score "red over yellow". "Greens" and "blues" were considered to be limited to technical roles and were explicitly not given opportunities to advance, though it took a long time to realize that. I started out thinking "hey, this is actually practical" and then over a few years went to "oh, they're using this to decide who moves up... That's a problem". I asked folks and was told by my manager's manager that ecolors were explicitly used in advancement criteria and who got opportunities to lead projects/etc. That's around the time I left. I hear they've dialed that particular bit back a lot, but it's still very weird to me that it's considered a normal and acceptable practice.

Mud isn't just to flush out cuttings. It's also to support the borehole. At large depths, you can't keep an open borehole and the "high energy" methods can't work because the borehole will collapse in on itself as it's created. Rocks are quite weak in tension. Drilling mud is dense and offsets the stresses trying to close to borehole as you get deeper.

I also learned from that book + tape about 30 years ago. I also have to strongly agree with this. It's quite different from learning stringed instruments. I'm sure it has a lot in common with some brass instruments, but either way, seeing doesn't help. Hearing and feeling is everything. That also makes it easier than most to learn via cassette tape! :)

The app seems neat, but focused on precision in bending. Precision in bends doesn't matter a ton for most styles. It's more about feel. Build and resolve tension. Don't worry too much about hitting things exactly. That advice will eventually fail you if you go into styles that need a more complete scale, but for blues/rock/country, it's much more about intuition than precision. The best way to learn is to play along with a lot of things and build "feel" instead of trying to precisely hit notes. (though you do need to hit a clean note to start bending)

We received a two knife set as a wedding gift many years ago. This was well before they were common, and it was quite a nice gift. I strongly agree with the "don't buy a knife you can't sharpen" point that folks often make, but ceramic knives do have their niche.

They're great for minimizing oxidation along cuts. E.g. cutting iceberg lettuce and avoiding having the edges turn brown. They're also very lightweight, which is nice for some things, while being bad for others. I'd never use them the way I do my workhorse chef's knife, but there are certainly tasks I prefer them for. Dicing lots of hot peppers comes to mind, oddly enough, as does some very precise and relatively tedious knifework like making very even matchstick cuts for carrots/radishes/etc (the large one has a very wide blade, which is great for this, and is lightweight enough to reduce fatigue).

Overall, I can see why folks like them. It's not really the "no need to sharpen" point. It's more the "lightweight and very thin" part, along with a non-oxidizing edge.

I'm still kind of opposed on principle, I suppose, but I do use the set we were gifted fairly often, despite having some very nice steel cutlery that I'm very fond of. I can't blame anyone for buying them now that they're priced more reasonably than they used to be.

Regular cameras respond to a wide range of wavelenghts, and they do actually reasonably mimic the response of the human eye.

Either way, it's the "range" vs "single wavelength" that's key here. The green band (or blue band or red band) isn't one wavelength. It's an average over a fairly broad range. Single-wavelength (or very narrow range) images are quite different.

You can model polygons on a sphere, but the issue is that the data you're starting with is already in a cartesian representation. You actually can't easily convert between the two for complex geometries in cases where they cross the antimeridian/poles. So trying to do anything other than points is difficult in practice, unless you're natively generating data from scratch in a spherical representation, which is rate.

For what it's worth, you _can't_ use spherical approaches for most data. They're only used for points, in practice. Your spatial data is inherently stored/generated in ways that don't allow spherical approaches as soon as you start working with polygons, let alone things like rasters.

Yes, spherical representations of polygon data exist, but the data you import has already been "split" and undoing that is often impossible, or at best non-trivial. And then rasters are fundamentally impossible to represent that way.

Analysis uses projections for that reason. Spherical approaches aren't fundamentally "better" for most use cases. They're only strictly better if everything you're working with is a point.

There's more to geospatial than point datasets.

>"I'd like to look like the guy on the right, but I'd rather be friends with the guy on the left"

This exactly. But I think the photo itself plays into that as well. It's not just a "safety" thing. (I'm not disagreeing that that's an aspect of it as well, though.)

The pose in the photo on the left comes off as more casual. "Hey, okay, I'm at the gym. Sure, let's do this before photo thing". The pose on the right comes off as "look at me! I'm showing off!". I mean, sure, nothing wrong with that in this context. I'd be proud too. But it's harder to relate to the person photographed as explicitly trying to be a show off than to the same person in a more casual (but still posed) position. The photograph on the left just comes off as a bit more humble and relatable just due to the pose / general vibe.

I'm really struggling to understand it as well. I mean, sure if what you're doing is a website, then maybe you can get something that functions out of an LLM. I don't really do web development, so maybe they're better for that specific niche.

However, for most cases I've tried, I get wildly incorrect and completely non-functional results. When they do "function", the code uses dangerously incorrect techniques and gives the wrong answer in ways you wouldn't notice unless you were familiar with the problem.

Maybe it's because I work in scientific computing, and there just aren't as many examples of our typical day to day problems out there, but I'm struggling to see how this is possible today...

It's called "mantle drip" or "lithospheric drip" in the technical literature. "Lithospheric mantle delamination" is a more descriptive variant. We're fond of fun terms in geology. E.g. the "jelly sandwich model" vs the "creme brulee model" for lithospheric strength is closely related to what's being discussed here. And yes, those are the proper technical terms.

Solids can indeed "drip". The mantle is a solid, but it still very much flows. "Fluid" means no shear strength, but fluids are not the only things that flow and flowing is separate from having a shear strength. Fluids do not behave elastically. If you stress them at all, they permanently deform. Elastic materials behave like a spring up until a point. When you stress them, they deform, but will pop back. The mantle is viscoelastic. Strain rate matters in how it deforms. When you stress at high strain rates it, it deforms, but pops back (i.e. shear waves from earthquakes can pass through). If you maintain those stresses at a low strain rate, it will slowly permanently deform. That's, by definition, flow.

This is a good analogy for what's being discussed: https://en.wikipedia.org/wiki/Pitch_drop_experiment

The reason "mantle drip" is used is to evoke a mental image similar to that experiment.

Just moved to Utah (wife's job was relocated here) a couple of years ago after many decades in the southeast and midwest. Some stuff grows like crazy here, some stuff doesn't.

Nothing in the continental US competes with the gulf coast when it comes to sheer masses of flowering vegetation. Everything is green and most of it flowers. Everywhere you look, there's actually really neat and beautiful plants. But most of it is what folks think of as "weeds". Non native stuff in yards (what people think of as "flowers") often struggles. E.g. yapon is native and is often an ornamental and it'll grow like crazy, or even worse, try planting some Mexican petunias... I love those things, but they are a purple swarm that will swallow everything.

But plant your daffodils, and while they'll grow, they get overpowered and outspread by everything native. A lot of common ornamentals can't take the fully saturated then dry cycles and clay soils, too. Then there are the bugs. Everything gets eaten by something. So so so so so many bugs.

But in Utah, those common ornamentals absolutely thrive _if_ they get water. When you irrigate here, things grow like crazy. Flowers are huge and there are basically no bugs here at all. (I don't count box alder beetles or brine flies as "lots of bugs".)

But native vegetation can survive the actual climate here, while most ornamentals can't without extra water.

It might help to think of it less as "Indian subcontinent slamming in" and more as the "Tethys ocean basin closing".

India being pulled away from Africa/Madagascar and eventually beneath the Eurasian continental crust is all related to the closing of the Tethys. ...This is a lot easier to explain with pictures...

If it helps, remember that continents are just lighter and thicker "rafts" of stuff. They're not plates themselves. They're often attached to oceanic crust as part of a plate. When one side of that oceanic crust becomes cold and thick enough to start to subduct, the entire plate gets pulled along.

That's a very personality and culture related phenomenon. A lot of folks will, but also a lot of folks won't.

Pointedly not playing the game when you have the political power to do so is often the most effective way to point out issues in the system that folks are being evaluated under. It can be a very wise move in some cases, as well.

CSV is _way_ hairier than folks think it is!!

And for anyone who's not convinced by CSV, consider parsing XML with a regex. "I don't need a full XML parser, I just need this little piece of data! Let's keep things lightweight. This can just be a regex..."

I've said it many times myself and been eventually burned by it each time. I'm not saying it's always wrong, but stop and think whether or not you can _really_ trust that "little piece of data" not to grow...

The big caveat is a big one. Choose your battles wisely!

There are plenty of things that look simpler than an established library at first glance (I/O of specialized formats comes to mind quickly). However, a lot of the complexity of that established library can wind up being edge cases that you actually _do_ care about, you just don't realize it yet.

It's easy to wind up blind to maintenance burden of "just a quick add to the in-house version" repeated over and over again until you wind up with something that has all of the complexities of the widely used library you were trying to avoid.

With that said, I still agree that it's good to write things from scratch and avoid complex dependencies where possible! I just think choosing the right cases to do so can be a bit of an art. It's a good one to hone.

^ This x1000. Everyone should read those! And the world needs more murderbot.

"As a heartless killing machine, I was a terrible failure."

“I was having an emotion, and I hate that.”

“I liked the imaginary people on the entertainment feed way more than I liked real ones, but you can’t have one without the other.”

And with that said, while I'm relatively pro deep sea mining, we simply don't know enough about deep marine ecosystems to balance risks. Every societal action has environmental impacts. You can't avoid them, and your existence in a technological society comes with a ton of deliberate environmental harm. What's important is to properly balance society's needs with the fundamental harm that our society's existence causes. Deep sea mining _could_ be a good way of doing that.

But we can't balance impact when we have so little information about the ecosystems at risk here. We just don't know enough.

Marine science as a whole is very expensive, but desperately needs continued funding and is critical for a variety of reasons. It's often said, but we really do know more about the surface of Mars than we know about the surface of our own planet. We have to keep funding ocean science for a huge variety of reasons.

I'm a marine geophysicist by background, just to give some context on why I'm starting this by griping about journals. Before I start trashing Nature Geoscience, I'd like to point out that this is an open access article and I give the broader Nature organization a lot of credit for their recent moves in that direction.

See https://www.nature.com/articles/s41561-024-01480-8 for the paper. It's quite short, and very much a geochemistry and/or geomicrobiology paper, so I am not the best person to really comment on it.

Now for a bit of ragging on the journal... The fact that this is in Nature Geoscience says a lot. I have an abysmally low opinion of that journal after seeing the first crop of papers published in it (it's fairly new). Lots of things that just didn't pass basic sanity checks of "is this physically plausible?". Hopefully it's improved... So far it's been "here's all the wacky crap that got rejected from Nature and couldn't be published in a reputable domain-specific journal".

Right now I treat it as a journal that has no clue how to pick peer reviewers. Therefore, stuff gets published that really shouldn't make it through peer review. While this _probably_ isn't one of those papers, I have trouble trusting Nature Geoscience to have picked relevant reviewers.

Despite having worked a lot on related topics, I don't know enough about deep marine ecosystems to really peel things back. With that said, we know there are plenty of different microorganisms that can produce oxygen in these circumstances. It seems odd to me to dismiss biological production immediately. Even if these samples were fully "poisoned" and that really did kill all microbes, why would electrolysis be the dominant source of oxygen for the entire ecosystem? Microbes are a lot easier to explain... But I am likely missing something critical there.

That's a key part of why I I'm focusing on the journal it's in. Based on previous publications, I have trouble trusting Nature Geoscience to have picked good peer reviewers on this. If it holds up, it should have been in Nature. I suspect it was rejected from Nature for some fairly fundamental reasons and then got through much lighter reivew in Nature Geoscience...

But I hope I'm wrong!

There are a lot of zany ideas that are unlikely but also aren't impossible. Those are pretty common and quite fun. Occasionally, they're correct. Much more frequently, they're not. But they make for a lot of interesting discussion and are incredibly important science. I hope this fits into that category.

I'm sorely tempted to ask a friend of mine who works on these exact types of microbes what she thinks of this... It'll make for an interesting discussion at least!

Even if it's homogenous, you have steadily increasing stresses with depth. This occurs for perfectly uniform materials as well. If you can't offset these to keep the hole open, it collapses.

Rocks are very weak in tension, despite being very strong in contraction. It's the reason you can break rock with a hammer or the reason ancient quarries were able to work by pouring water on wood pegs in rocks. It's also the reason concrete needs rebar to reinforce it (steel is very strong in tension, so the two combined are exceptionally strong). Keeping a hole open requires strength in tension as well as strength in compression.

Drilling mud accomplishes this by being roughly the same density as the rock, so it offsets the stresses that are trying to close the borehole that steadily increase with depth due to the increasing amount of rock above. Drilling mud keeps the borehole open until you can put in casing to support it.

This is exactly the same reason why it's difficult to build a submarine that can go to very large depths in the ocean. To put up steel walls (casing) to keep it open, you have to stop drilling and cement in casing - you can't do that as you drill. So drilling mud is a key part of being able to drill efficiently. Otherwise, you'd need to stop every few tens of meters and spend _days_ setting casing before being able to drill again.

Regardless, there is nowhere on earth where things are homogenous over very long distances. Simply put, even relatively uniform rocks can have very significant variations in physical properties. Many relevant properties (e.g. permeability - how well fluids can move through) vary over _tens of orders of magnitude_ naturally. So "uniform" can still mean "only varies by a few orders of magnitude". There are places where you can reasonably avoid non-silicates, but you're going to hit tons of other issues due to fundamental heterogeneity.

If you're near a port of entry, you actually can be stopped at any time to be asked for proof of citizenship/visa/etc. See also https://www.aclu.org/know-your-rights/border-zone There are plenty of border patrol stops along highways that never cross a border where you'll be routinely stopped.

In practice, that means "near the border with Mexico" where there are border patrol stations along highways well inside the US that do routine stops, but technically it applies to any port city as well.

_Technically_, if you're a US citizen, you're not required to have proof of citizenship on you. But a drivers license in the US does not show that you are a citizen, and they absolutely can stop you from leaving (at least long enough to mess up your day plans) if you cannot actually prove citizenship in practice. It's just a _lot_ easier to have a passport on you. Passport cards exist mostly for this purpose.

E.g. I lived in Texas for a long time and have a passport card for that specific reason. Driving down to the beach or just going on random trips involves passing through border control stations even though you never leave the US or even get particularly near the border. Your chances of just going to visit a friend in south Texas and winding up needing to pass through a border control checkpoint due to unplanned road closures/etc are pretty high, so it's best to have a card in your wallet at all times that counts. Most remote highways have them as soon as you get back near a town. Most of the time, they'll wave you through pretty quickly, but it's just a lot faster process if you have a passport card and not only a drivers license.

Side note: A friend of mine was a park ranger at Carlsbad Caverns. He picked me up at the El Paso airport and we were driving to Carlsbad. We went through a border control checkpoint. They threatened to arrest him on the spot because he had his park service uniform hanging up in the back, but no employee ID or other proof that he actually was a ranger. They said he was impersonating a federal officer. They made a big deal of it and made us wait while they called the park to verify he was employed (and that didn't work because it was about 8pm). We eventually got let go with a warning after he found random online photos of him giving cave tours and convinced them that he couldn't possibly have staged a ton of random folks on the internet to have him in the background of photos.... But still, it was kinda nuts...