Full Width 
Shortcuts 
Sign Up 
Log In 
Articles 
Reading List More from Rubenerd
I feel like I righted a wrong in some small way recently. It also gives me an excuse to write another of my shaggy dog posts about old computers, which I’ll never pass up the opportunity to do. Behold, our boxed copy of Digital Research DOS 6.0 from 1992: The first OS I ever used was MS-DOS 5.x on the family 486, upgraded shortly after to MS-DOS 6.2 [sic], then Windows 95. DOS was “what came with computers” to me at the time, so the idea there were alternatives didn’t even cross my mind. MS-DOS’s origins lay with QDOS, itself a broad clone of Digital Research’s CP/M operating system. CP/M-86 was compiled and offered for the original PC, but IBM famously went with Microsoft instead. The circumstances surrounding this decision remain… contentious. Either way, I’ve had a soft spot for Gary Kildall since I first saw him on those archived episodes of the Computer Chronicles. People like exploring an alternative universe where IBM went with CP/M over DOS, and Gary had become the billionaire tyrant instead of Gates. We can only speculate, but Gary always struck me as more of a self aware, Woz-type character than Bill. I’ve mentioned how Digital Research’s GEM desktop was a big part of the reason why I got into the Atari ST, but I haven’t talked much about DR DOS, the OS Digital Research released to compete with Microsoft and IBM. I never used it as a kid, but I wondered what Gary and his team had been able to deliver. Last year I acquired disk images for every major release, in part for a blog post series about DOS that I never ended up publishing. Version 6.0 quickly supplanted IBM PC DOS 7 as my preferred DOS; everything from the installer to the included utilities are vastly superior, and it even came with ViewMAX, a file manager based on GEM/3. Awesomesause! Pair this with Turbo Pascal, Commander Keen, Sim City 2000, The Games: Winter Challenge, the Brief editor, and the DOS port of Lemmings, and I’ve got all the bases covered! DR DOS 6 was released in 1992, around the time Digital Research was acquired by Novell. For context, Lynne and William Jolitz released 386BSD at the same time, and Microsoft was selling MS-DOS 5.0 to my dad for our first family PC. He could have ended up with DR DOS 6.0 instead, which makes this all the more interesting. This boxed set of DR DOS 6 includes the instruction manuals for DOS and ViewMAX, along with a handy reference card. The “Value Pack” sleeve contained the install disks, but I’m not sure if it previously had something else. The OS itself is contained on two 5.25-inch high density disks, along with the original voucher for 360 KiB disk versions. So many of the disk labels from the time seem to have developed this blotchy appearance, but fortunately it’s only cosmetic. Funnily enough, my PC DOS 2.1 disk is about the only one from the early 1980s that hasn’t developed this odd label issue. In the next post, I’ll get a basic install of DR DOS 6 going, to demonstrate why I like it so much. Only took me (checks calendar) two and half decades. By Ruben Schade in Sydney, 2026-09-02.
Snow Leopard was a version of Mac OS X released in 2010. It was billed by Apple as having “0 new features”, with a focus on refining the system, improving efficiency, and fixing bugs. Those of us using Macs at the time welcomed the news after living with updates that seemingly broke as many things as they introduced with each release. This idea is so powerful—and so longed for—that it’s escaped containment among the Apple crowd. I’ve seen everything from Linux distro to phone updates referred to as Snow Leopard releases, when their vendors cite stability and bug fixes over new features. Likewise, people plead with their vendors for a Snow Leopard release when they feel quality has slipped. The reality at the time was a bit different. As I wrote at the time: This could be the first time I’ve ever downgraded an Apple OS. As we speak I’m downgrading the Snow Leopard partition on my trusty old Core Duo 2006 MacBook Pro back to Leopard (and while I’m at it, I’m finally getting around to upgrading the FreeBSD partition to 8.0!). I kept waiting for updates to fix the severe stability problems in the Finder, my FireWire 800 ExpressCard and my iMovie HD plugins, but I’ve decided to give up for now. This was putting it mildly. When I tried it again a few months later, I still had so many problems that I downgraded again, then leapfrogged it for Lion when it was released in 2011. Jeff Johnson has a great post discussing all the problems Snow Leopard introduced and had to have patched. It was a mess. Still, whether Snow Leopard really was this bastion of stability and polish, the myth surrounding its development and stated purpose obviously resonates today. People invoke its memory because they’re broadly craving stability and quality of life improvements, something almost everyone has wilfully discarded as they chase the next rung of enshittification. Apple’s marketing team hit on something so popular that it has staying power sixteen years later. By Ruben Schade in Sydney, 2026-08-02.
I worked briefly in process control when I was younger, and saw Things.™ One of those Things™ was a wall of LEDs to indicate statuses of various valves and sensors across the entire plant. It was old school, but it worked. It also made you feel like you were on a 1960s-era sci-fi set. What does this have to do with databases? That’s an extremely excellent, and dare I say handsome, question. While we’re on the topic of things I used to do, I used to work against a large database that had a schema with a specific quirk that used to drive me up the wall. Take this contrived example of a table for an account: username | active | paid ---------------+--------+------ nagato_yuki | true | true takarada_rikka | true | false And another for some basic infrastructure: hostname | running | error ------------------+---------+------- retro.rubenerd.au | true | false www.sasara.moe | true | true Do you spot the subtle difference between the two? Why is the first table well designed, and the second one horrible? The first table has uniform positive attributes. Having an active account is good, and being paid is good. This means one can tell at a glance whether an account is in good standing or not. You can also do trivial boolean algebra to derive a status. The second table has a mix of positive and negative attributes. Having a running state of true is good, but having an error state of true is bad. This means you can’t as easily tell at a glance the status of an instance. You can’t also as easily derive an overall status. The value of consistent booleans was something that was immediately obvious when looking at that status wall. It mostly had white lights. When one of these changed colour, you knew there was something actionable. If that status wall instead was designed with some lights being positive, and some being negative, it would be almost (but perhaps not entirely) useless. ☕︎ ☕︎ ☕︎ Why am I thinking about this today? I’ll admit it’s for an incredibly embarrassing reason: I had a dream about databases last night. I know, some people have fun dreams involving flight, reliving something they enjoyed, or exploring a new side of themselves. I dreamed I was back at uni with my DB professor who insisted that all booleans should have positive assertions. Anyway, I was shy at the time when I went to that class in the real world, and said nothing. But what I wanted to say was that uniformity is more important. Most of the time there’s an unambiguous, convenient term to describe something positive, but I also acknowledge there are some circumstances where forcing this naming can get clumsy, such as having a schema full of not_$foo attributes. It’s not a data science to me, but I’m reminded of the sixth rule from The Economist Style Guide: Break any of these rules sooner than say anything outright barbarous. The same applies for software variables, but that would be a whole other post, can of worms, and kettle of fish. By Ruben Schade in Sydney, 2026-07-01.
So here’s the thing. Late last year I bought the excellent SilverStone FLP02, a retro-themed computer case with modern component support. It was absolutely, jaw-droppingly gorgeous, and has easily become my favourite modern case alongside the legendary NCASE M1. It’s big, relatively easy to work in with just a few caveats (a topic for another time), and yet looks the part so well. I’ve since moved our FreeBSD bhyve/jail server into our FLP02, and am giving serious consideration to an additional one for my desktop in the loungeroom. It’s that good! I subsequently received email from angry people asking me what the point was, which I answered in a detailed reply. Turns out that just made them more cranky, because they thought I was mocking them somehow. Perish the thought! But use of such a case poses a problem, and it’s the same one the plagued seemingly all beige computer towers at the tail end of what I call the Golden Era of computer cases. Which I suppose should really be called the Beige Era. The era when computer case manufacturers thought the buying public should be able to buy whatever colour they want, as long as it matches a Model T Ford. It was such a disappointing development. If we were going to replace beige—a colour mostly unique to computers in the consumer electronics space—why did it have to be the most boring colour (or lack thereof) of all time? But I digress. The biggest problem beige towers face is the pairing of drives and peripherals. During that transition period between awesome beige and mediocre black, you were lucky to get a part that matched. Black optical drives in beige cases, and beige disk drives in black cases became a distressingly common sight. Retrospectively, I have friends who harbour a nostalgic attachment to that anti-design. To them, a beige case doesn’t look complete if it doesn’t have a mismatched drive. I respect that position, but it’s not for me. I grew up with beige DIY towers with beige drives, and that’s what I want for this glorious FLP02 damn it! I want stuff to match, even though I concede plenty of people mismatched components owing to price, convenience, and/or availability. Which leads us to this contraption pictured here: This is a drive cage, designed to slot into two 5.25-inch drive bays and hold three 3.5-inch hard drives. While the FLP02 is stupendously wonderful, it technically has fewer internal drive bays for storage than the Antek Three Hundred I used to use, which resulted in me needed to purchase additional drive capacity. This drive cage allows me to mount an 80 mm Noctua fan (itself beige!) into the front, and keep three additional drives cool. I have a separate post about how this works pending. I prefer finding beige components when I can for new and old builds alike, but when that isn’t possible I’ve developed a system that (mostly) works using these three cans: They are, from left to right: A self-etching primer. I like the White Knight Rust Guard SLS the most, though I’ve had success with others too. Primer gives your paint better adhesion, and self-etching means you don’t need to do any manual work to rough up the surface to accept the paint. This is especially important for smooth painted surfaces or bare metal, though it’s also worked great on rougher powder-coated computer parts for me. A colour, of course. Rust-Oleum Chalked Chiffon Cream is brighter than the beige used in computer components, but it’s the closest match I’ve found thus far after exhaustive testing. A top coat to protect the painted surface. I like Rust-Oleum’s Satin Clear, becuase it gives a similar finish to the powder-coated surfaces on beige powder coatings. Painting takes a while, but I’ve found being slow and building up very thin coats works best. My approach: Wipe down and clean the item to paint with isopropyl alcohol until there isn’t any dust, fluff, or other detritus across the surface. If you don’t, and you paint over the top, you’ll get a lump that will have to be sanded down, and likely result in you repainting the whole thing. Place a tarp down over a table or other surface outside, and weigh it down with some heavy objects. We use a few large stone coasters. Don a mask too; you don’t need to vape paint, let alone anything else. Place the item to be painted on top of the tarp. Shake the cans vigorously for at least a minute. In cold weather, use a hair dryer to heat the can and its contents slightly (though obviously not too long). This is a tip multiple artists and industrial designers have told me about, before you write an email telling me this is a bad idea. Hold your can about 20-30 cm away from the part, and spray across the surface with your primer, overlapping a little with each pass. I find two coats works best, with at least an hour between each coat. The cans claim they dry in less than ten minutes, but yeah, nah. Spray three to four thin coats of your beige coloured analogue, leaving at minimum an hour in between again. Err on the side of not applying enough instead of applying too much. You can always build up more layers, but if you apply too much at once, it’ll goop up, take forever to dry, and will get stuff stuck to it. Every mistake I’ve made has been down to impatience. Finish it off with one or two coats of clear satin finish. This is what my drive case part looked like after two coats of the coloured paint. You can see the lower edge is slightly darker than the rest, so I went over that area a bit more on a subsequent pass. Note also the droplets of rain on the tarp, which required me to pause painting last week (cough). Once I’d finished, I attached it back to the drive cage and installed it into the FLP02, along with the Noctua fan and the supplied dust filter. Success! As you can tell, the colour isn’t exactly the right shade of beige, but it’s still a significant improvement (to me) over the matte black of the previous part. If anything, slightly yellowed case parts look closer to it, so this approach works especially well for older towers. Now I just need a SATA DVD-RW in the appropriate beige colour, and another drive sled to arrive from SilverStone (post incoming!), and this build will be done. What an awesome case. What a fun time! By Ruben Schade in Sydney, 2026-06-02.
More in life
This is a weird moment. I have a hunch that we’ll look back later and tell ourselves stories in which it seemed obvious where things were going. So I wanted to lay down a little time capsule and register just how confusing everything felt at the time. This is a personal view from the trenches. 1. If you’re a normal person, you may feel that AI has been good at math for a long time.1 But for “real” math, this is roughly what it felt like: When Feeling 2024 LLMs are toys Summer 2025 LLMs are interesting toys Early Autumn 2025 LLMs can do useful stuff sometimes Mid Autumn 2025 …a lot of stuff Late Autumn 2025 …a lot Early 2026 holy god Even having been astonished by AI progress in the past, it was still astonishing to see AI move in a matter of months from a mildly-useful toy to something that many people now consider almost impossible to do without. 2. Personally, I’d say that current AI makes me around 50% more productive at “doing math”. That’s despite the fact that AI can now often do in a few minutes things that would have taken me hours to do myself. The main reason is that my concept of “doing math” involves various steps, like coming up with ideas for theorems, trying to prove them, writing things down, reorganizing everything to make it more clear, and so on. AI is good at some parts of this, but still pretty bad at others, so the overall speedup isn’t as high as you might expect. (Even if planes could go 10× faster, your door-to-door travel time from your apartment in London to your friend’s place in Rome wouldn’t decrease that much, because you still have to get to the airport and wait to board.) 3. But then, I’m still (sort of; more on this later) trying to follow my old “doing math” workflow, just automating certain parts of it. That can’t be right, can it? If we found a way to make batteries at 1% of the current cost, we wouldn’t just say, “cool, batteries are cheap now”. We’d find all sorts of new ways to use batteries that didn’t make sense before, and change the ways we generate power to account for all these new batteries. Probably our whole energy system would end up transformed. Analogously, AI being good at proofs should lead to a very different equilibrium. But what is that?2 4. It seems healthy, at the moment, to maintain a fluid view of what “doing math” means. But at bare minimum it surely involves understanding the math that “I” am “doing”. So long as that’s part of the workflow, it would seem to impose a limit on how quickly “I” could “do” math, no matter how good AI is. One obvious conclusion would be that my tiny organic brain should be removed from the workflow entirely. Maybe! But it’s interesting to speculate about how much faster things could go while still keeping a human brain in the loop. I think the answer is: A lot. 5. Contrary to what you may have heard, humans are not obsolete for math. Not yet. The most obvious reason is that AI is still bad at writing. People often complain that AI writing overuses various patterns or lacks a unique voice or whatever. I don’t disagree, but I always felt like there was a deeper issue that I couldn’t quite put my finger on. With math, this becomes very clear: The problem is it doesn’t make any goddamned sense. Really, I think it’s deeper than being bad “at writing”. It seems like AI is bad at thinking in certain ways that manifest as bad writing. 6. When novices do math, they often create “spaghetti-code math”: Imagine a long document with lots of equations, but if you point at any one equation, it’s hard to say what assumptions underpin it, or how it fits into the bigger picture. Any random idea that was mentioned 17 pages ago might be used again at any time with no warning. With experience, people learn to break complex arguments down into modular pieces with a good “API”, in the sense that the assumptions and conclusions for each part are relatively simple. Somehow, AI has developed these amazing technical skills, while still being bad at creating abstractions So the arguments it gives are long and tangled and bewildering. This is what makes AI math seem so alien. 7. Now I know what you’re thinking. Maybe AI isn’t bad at creating abstractions. It’s just bad at creating abstractions that work for you, human with tiny organic brain. From a distance, that seems plausible. And it might be true to some degree. I have high-bandwidth access to a human brain. It’s not crazy to imagine that this gives me an advantage in predicting how a human brain would react to a given presentation of ideas. But still, I claim that current AI is objectively bad at creating abstractions, regardless of if the consumer of those abstractions is human or artificial. Here’s why: If I ask AI to prove something hard, it will produce a proof that starts simple and gradually descends into a rampage of complexity. If I tell it to break that proof down into self-contained pieces, it will produce lemmas that each take a full page just to state. If I ask, “Why? Why have you done this to me?”, the AI will say, “This is as simple as it gets. It’s not that bad. What’s your problem?” But if I argue with the AI for a while, I can start to understand the basic contours of the argument. And I’ll ask, “Why don’t you first just try to prove blah using only assumptions blahblah, without worrying about blahblahblah? Wouldn’t that be easier?” Often, this doesn’t work, either because my idea was bad, or because the AI isn’t smart enough to do it. (Which one? Who knows!) But after a few attempts, I can usually find something that that works, and the AI can use it to re-write the proof in a somewhat simpler way. If I repeat this many times, I can gradually chip away at the complexity. But maybe that’s still just making the proof easier for humans to understand? No! For one, this usually makes the proof shorter, and sometimes dramatically shorter. More importantly, during this process, someone (maybe me, maybe the AI) will often realize that there’s a different proof strategy that gives a stronger result. Instead of proving x < 9.7, now we can prove x < 7.1. Thus, artisanal human abstractions are “objectively” useful. 8. Some people might claim that it’s already possible to automate some of what I’m doing with some multi-agent Rube Goldberg thing. I’m not sure, because my experience has been that investing lots of time into clever AGENTS.MD or custom skills or harnesses isn’t a great use of time. That stuff is somewhat helpful, but things are changing so fast that the knowledge has a short half-life. Also, the majority of information about these tricks seems to come from people who are excited about the technology itself. On the other hand, the automator told me to try typing, “ask independent agents for an adversarial review”, and I’ve found those seven words to be quite helpful. I wish I could find a source with boring best practices for people who are happy to stay away from the bleeding edge. 9. Another reason AI doesn’t improve my productivity more is that it is often the green light, the orgastic future that recedes before me. I’ll ask AI to do something, and it will almost seem to do it. So I’ll ask for changes, and it seems to help, but it’s still not quite right. Hours later, I’ve accomplished nothing and learned nothing. Eventually, I start over, either doing it myself or with AI on a very tight leash, and I’m done in 30 minutes. Each time this happens I feel foolish and tell myself not to do it again. But it’s very hard to judge in real-time. 10. Of course, I’m not claiming that AI is intrinsically bad at writing or thinking or abstraction-making. It’s gotten better at all of these and I expect it will continue to do so. All I’m saying is that as of Autumn 2026, humans still have a lot to contribute, and it’s not just better exposition. 11. Did AI actually get much better at math in Autumn 2025? Or perhaps, in the grand scheme of things, human math ability falls into a narrow range? 12. Some people are responding to the new reality by gleefully using AI to generate lots of semi-OK-ish papers and puking them up on arXiv. I understand the incentives that drive this. In some sense, it seems like the logical response to the new reality. (After inventing cars, we started taking a lot more low-value trips.) But I don’t think this is socially useful, and I tend to think everyone will soon agree that this is embarrassing. But… I’m not sure. Maybe it is socially useful? And I find myself feeling vaguely jealous of the people who are wired this way, who instead of feeling blinkered, see opportunity and plunder. 13. Want an optimistic pro-AI, pro-progress take? OK. If I take my old “doing math” workflow, AI makes it strictly better. A lot better. It’s like having a hyper-conscientious intern that is smarter than me in many ways, works at incredible speed, and knows approximately everything. Yes, AI sucks at writing and making abstractions. But if I’m disciplined and try to mostly do the math “myself” while asking the AI for help, I can explore ideas much faster than before, make far fewer mistakes, and produce something just as clear as before, while developing just as much understanding. I can ask, “What happens if you do a Kofuni expansion?”, and get an answer instantly. I can ask for illustrations or interactive visualizations and see them instantly. Arguably, we’re incredibly lucky to be alive at this moment, when these new tools have opened up this new territory to explore. 14. And isn’t that what I wanted? Back in 2024, I bemoaned that what AI most seemed to struggle with was getting the facts straight: Perhaps the old science fiction robots were a projection of our fantasies. I want my style and voice to be precious and uncopyable. I want an AI assistant that will tirelessly hunt down facts while I provide creativity and panache and magic. But that’s not how it is. So we should remember: In a scenario where AI is useful but humans retain some comparative advantage, we might not like where that comparative advantage lies. At least with math, and at least for the moment, I get to be Kirk, while the AI is Spock. (Even with other types of writing, my impression is that since 2024, AI has improved more along Spock dimensions than along Kirk dimensions.) So that’s nice. 15. In the last few months, it frequently happens that a collaborator will use AI to generate something, and then send it to me and ask me to check if it’s useful. I can see why this seems reasonable. (Really, I do.) But when it happens a lot, one can’t help but wonder what the point of the human collaborator is. 16. On the other hand—with creative writing, my advice is not to use AI to write, at all. Ask questions if you want, but write every damned word yourself.3 With math, I don’t feel like such a hard line makes sense. It seems that creating something good still requires a lot of human involvement, to the point of agonizing over notation, repeatedly re-organizing things, and heavily editing individual sentences. Still, I find that if people make that effort, they can create something just as good as they could by hand, while still being perhaps 50% AI written.4 So I don’t mind if people send me “centaur math”, provided they take responsibility for the contents. The problem is that when working with AI, it takes constant paranoid vigilance to avoid letting things slip beyond your understanding. I often find myself wanting to awkwardly ask people, “OK, you obviously used AI to make this… which is OK… but how much of this do you vouch for? If we have a meeting to discuss it, will you crumble the moment I start asking questions?” 17. When I’m in abstraction-creation mode, I feel like I’m doing something intelligent. When yell at AI about bad notation and writing, I feel like I’m providing taste. But I often find myself giving feedback that seems incredibly basic. Things like: This is a mess, clean it up. Or: Do you need that assumption? Or: If you can’t solve this problem, why don’t you create a simpler version and try to solve it first? On the one hand, I guess a lot of what good human managers do amounts to giving very basic feedback. On the other hand, it’s hard to imagine that I have much durable advantage in doing this kind of thing. So I wonder how much durable advantage I have in abstraction-making or taste, too. 18. A common reaction to AI getting better at proofs is that humans need to focus more on exposition. This has an obvious Bourdieu-esque criticism: Human painters used to strongly value photo-realism, but after cameras were invented they convinced everyone that, oops, photo-realism isn’t important, what matters is expression, which coincidentally cameras can do. Isn’t a shift towards exposition in math just a way for humans to move the goalposts and justify their own continued relevance? The only credible way I see to avoid that criticism seems to be to bite the bullet and say that, in fact, math should have already have focused more on exposition, even before AI. Personally, I strongly believe that this is true. 19. I think the biggest reason I’m not more productive is: What’s the point? In the past, if you came up with an elegant proof, you’d tell yourself, “I made a thing! Other people will read and use my thing! I am advancing human progress.” You can still tell yourself that now. But you have to wonder, will anyone actually read your proof? Or will they just ask another AI to explain it to them? Will your proof help their AI explain it better? For how long? If you just waited a year, would some other AI produce that same proof independently? Alternatively, if you’re a normal person, you might think that LLMs still struggle with basic arithmetic? ↩ AI progress is so fast, compared to society’s ability to adjust, that we may never actually see what “the current equilibrium” looks like. ↩ This essay, if you’re wondering, was created by writing every damned word myself and not asking AI anything. ↩ You could have a whole debate about what that precentage means. It’s not the number of words written by AI that really matters, in my opinion, so much as who is responsible for the overall structure of the ideas. ↩
I always liked stories about eccentric royals doing mad things. Queen Christina of Sweden reportedly had a miniature cannon made for firing tiny cannonballs at fleas in her bedroom. Louis XIV made a proto-reality-show of his daily routines, allowing an audience of courtiers the honor of watching him dress, bathe, and use the toilet. Peter the Great famously returned from […]
The (Re)insurance AI Landscape
to become the version of yourself needed in the next chapter
