Diag| Considering item [tag:pappp.net,2026-08-07:/2502960] "2027 memory capacity is reportedly sold out"
Diag| Considering item [tag:pappp.net,2026-08-05:/2502506] "Sycophantic AI Decreases Prosocial Intentions and Promotes Dependence (2025)"
Diag| Considering item [tag:pappp.net,2026-08-04:/2502146] "All of Winona Police Department's Flock cameras cut down and stolen"
Diag| Considering item [tag:pappp.net,2026-08-03:/2501883] "Celebrating 45 Years of Kermit with the First New C-Kermit Release in 15 Years"
Diag| Considering item [tag:pappp.net,2026-08-02:/2501525] "MkLinux and the pimped-out Apple Workgroup Server 9150"
Diag| Considering item [tag:pappp.net,2026-08-02:/2501516] "ASRock BC-250: Building the Budget Steam Machine"
Diag| Considering item [tag:pappp.net,2026-07-31:/2501295] "Google plans to exempt sanctioned nations from Android developer verification"
Diag| Considering item [tag:pappp.net,2026-07-28:/2500259] "College prof hides prompt to catch AI cheaters, finds human nature is pretty much as we thought"
Diag| Considering item [tag:pappp.net,2026-07-28:/2500268] "Now is the time to give LLMs access to the ACM digital library"
Diag| Considering item [tag:pappp.net,2026-07-22:/2498801] "John C. Dvorak has died"
Diag| Considering item [tag:pappp.net,2026-07-22:/2498669] "Passkeys were invented by engineers with zero understanding of consumer brain"
Diag| Considering item [tag:pappp.net,2026-07-19:/2497943] "Automated Pressure Advance Using a Bed-Leveling Sensor"
Diag| Considering item [tag:pappp.net,2026-07-19:/2497945] "Building an Arch Linux Aarch64 Port for Holo Core"
Diag| Considering item [tag:pappp.net,2026-06-18:/2488796] "What was nice about the UI of Windows 2000"
Diag| Considering item [tag:pappp.net,2026-06-17:/2488496] "Lore – Open source version control system designed for scalability"
Diag| Considering item [tag:pappp.net,2026-06-16:/2488199] "Linux kernel 7.1 sends Intel 486 support to silicon heaven"
Diag| Considering item [tag:pappp.net,2026-06-15:/2487839] "Fox is buying Roku"
Diag| Considering item [tag:pappp.net,2026-06-14:/2487662] "Any road followed precisely to its end leads precisely nowhere."
Diag| Considering item [tag:pappp.net,2026-06-19:/2489115] "I used sound waves to make espresso"
Diag| Considering item [tag:pappp.net,2026-06-10:/2486637] "German court rules Google is liable for whatever Google’s “AI” generates"
Diag| Considering item [tag:pappp.net,2026-06-01:/2484373] "Microsoft is intentionally bricking all Office for Mac 2019/2021 installations"
Diag| Considering item [tag:pappp.net,2026-06-01:/2484272] "Nvidia's Grace Blackwell superchips are officially coming to the PC with RTX Spark notebooks"
Diag| Considering item [tag:pappp.net,2026-05-30:/2483921] "Adding Linux support back for the BASIC (free) version of Vivado"
Diag| Considering item [tag:pappp.net,2026-05-30:/2483835] "Proposed new US funding rules: We can cancel any grant at any time"
Diag| Considering item [tag:pappp.net,2026-05-27:/2483228] "The exemptions in age-verification laws for open source operating systems are bad, actually"
Diag| Considering item [tag:pappp.net,2026-05-31:/2484064] "Mysteries of the Griffin iMate"
Diag| Considering item [tag:pappp.net,2026-05-23:/2482255] "AMD (Xilinx) is Excluding Linux From the Free Tier For Its FPGA Dev Tool"
Diag| Considering item [tag:pappp.net,2026-05-19:/2481210] "The Virtual OS Museum"
Diag| Considering item [tag:pappp.net,2026-05-19:/2481121] "Google changes its search box"
Diag| Considering item [tag:pappp.net,2026-05-15:/2480224] "Windows 11 tests an adjustable taskbar and resizable Start menu"
Diag| Considering item [tag:pappp.net,2026-05-15:/2480167] "Send the arXiv AI-generated slop, get a yearlong vacation from submissions"
Diag| Considering item [tag:pappp.net,2026-05-13:/2479471] "KDE Receives $1.4 Million Investment From Sovereign Tech Fund"
Diag| Considering item [tag:pappp.net,2026-05-12:/2479155] "Google's Android-powered laptops are called Googlebooks, and they're coming this year"
Diag| Considering item [tag:pappp.net,2026-05-07:/2478063] "Canvas is online again after ShinyHunters threaten to leak schools’ data"
Diag| Considering item [tag:pappp.net,2026-05-07:/2477845] "Aramark, University of Kentucky to end partnership, eliminating more than 900 jobs"
Diag| Considering item [tag:pappp.net,2026-04-29:/2475951] "Apple gives up on Vision Pro, disbands Vision Pro team"
Playing with the Raspberry Pi has me thinking about it’s touted promise as an Educational Computer, one of its stated design goals. I actually think about what a good computer for learning about computers should be quite a bit, out of both personal and professional interest, and have been building a list. I hear an older generation talk about their Apple IIs and ZX Spectrums (and occasionally even access to mainframes through parents) as formative experiences, and I’ve come to appreciate that the stream of interesting old computers my parents kept me supplied with as a child were an amazingly formative experience, and one I can never thank them enough for, so it is interesting to think about what will fill this niche for children growing up now.
For the sake of brevity, we’re going to call this kind of computer a “Learning Computer.” My idea of a good learning computer is:
A Real Computer
A good learning computer has to be usable for the same kinds of tasks as contemporary computers, in the same general manner. It can be limited and inferior, but it has to be basically analogous. Little development boards and Microcontrollers and the like usually don’t make it on this point.
Not “The” Computer
A good learning computer will always be a second (or n-th) machine. No one wants to break the computer they use day-to-day, and learning on a machine people depend on restricts one’s ability to really tinker with it. It is especially important to have another machine with which to get help when things go wrong with the one you are tinkering with.
Cheap
A good learning computer will be both cheap enough to be given to or purchased by anyone curious, and cheap enough that the owner isn’t overly afraid of destroying it. This is fundamentally the same issue as the previous point.
Common, but not Ubiquitous
Truly obscure machines will lack documentation or community, so they will be hard to learn with. A good learning computer will be sufficiently widely distributed that there will be resources and a community available for it. Conversely, for truly ubiquitous computers there will be a canned solution for any problem, so one has to go out of your way to learn anything from it. There is even some value in it being a little bit weird, so that the owner can appreciate design distinctions.
Open
You need to be able to reach in and poke around at the software and hardware. Ideally, the entire boot process will be visible, and at least one operating system the hardware will run will be capable of introspection. The level of openness required is negotiable, but an iOS gadget isn’t going to do it.
Flexible
A good learning computer should be able to boot multiple environments, so that the owner can experience a variety of systems, and appreciate what is dictated by the hardware and what is not. It should also have hardware subject to tampering, swapping, and replacement for the same reason. Swappable storage media is especially important on this front.
Programmable
There are few things more powerful than the knowledge that one can make real things. A featureful, accessible programming environment is perhaps the safest, cheapest way of imparting that knowledge.
Just to enumerate how a Raspberry Pi stacks up –
It hosts a variety of environments that look and feel much like a full grown computer, and also is deeply similar to the baby computers / overgrown appliances that have recently become common. It costs $35, and isn’t really pitched as a primary computer. It has already sold enough units to be common, documentation is widely available, and a community has sprung up among the hackers and makers, but it isn’t oversaturated. The default software stack, with some really disheartening holes, is basically open and inspectable. It’s modular via USB and SD cards, and is as accessible as anything with BGA packed parts can be. It is programmable in almost any common language, and many environments floating around for it are tailored to that use.
For the time being it is hard to argue against an old PC, but PCs are increasingly becoming blackboxes that only speak complicated protocols, and older simpler PC hardware is quickly falling out of support even in the hobbyist OSes, which carves a clear niche for a purpose-made learning computer like the Raspberry Pi. It really is very close to the machine the upcoming generation needs as a learning computer, but it remains to be seen how well it will do at reaching the people who will be enriched by it.
This came out of a discussion on plus about an article I shared, but Plus isn’t really a suitable venue for the long-form response. The basic premise of the article, like so many similar articles, is that we need to impose continuous evaluation, and tie incentives to the results. A great many well-qualified people (preferred example: Diane Ravitch) think that model is wrong, and has already done a great deal of damage, and I tend to agree with them. (also note, the choices of some points and examples here are oriented for people with known similar experiences). This is also the first time I have set up this whole argument at once, so I’m sure there are holes you could drive a truck through in at least some of my claims, but I’d like to try for discussion’s sake. Continue reading →
Even if you do something that others might consider wrong, you should at least be willing to talk about it and tell your parents what you’re doing because you believe it’s right.