System Stack

You Can Have Colemak on DOS, Too

2026-05-20T00:00:00+00:00

If you've picked up a Pocket 8086—and if you're curious, my guest post for Stephen at 512 Pixels</a> is probably the place to start—you've got a genuine PC-XT in your pocket, one that comes running MS-DOS 6.22 out of the box. Is that kosher according to copyright laws and such? Should we be buying Chinese-made hobbyist computers with completely unlicensed software and probably "borrowed" open-source BIOS as well? I'm going to pretend you didn't ask those questions and continue with my blog post.

But let's say you actually want to use the thing for something, and let's also say that you taught yourself to touch-type Colemak</a> years ago because it helps you avoid RSI from being on a keyboard all day.

Starting point: the command line</h2>

If you're in DOSBox-X or something else running FreeDOS, you just type keyb co</code> and you're done. Welcome to the future.

For those of us running "real" MS-DOS, getting Colemak working at the DOS prompt is straightforward, if you have a way to transfer files onto the machine: grab KEYB.COM</code> and KEYBOARD.SYS</code> from the keyb</code> and keyb_lay</code> packages on the FreeDOS package archive at ibiblio.org</a>, drop them in C:\TOOLS</code> (or wherever you want, but this is where I put them), and add this line to your AUTOEXEC.BAT</code> so it loads at startup:

C:\TOOLS\KEYB co,,C:\TOOLS\KEYBOARD.SYS
</code></pre>
That gets you Colemak everywhere that DOS handles keyboard input. Great!</p>
The next level</h2>
The problem is that the serious word processors available on DOS don't let DOS handle keyboard input—they do it themselves. XyWrite</a> and WordPerfect</a> bypass the DOS layer entirely and talk to the hardware directly, which means KEYB</code> does nothing for you inside them. Each application has its own keyboard definition format, and you have to configure Colemak separately for each one.</p>
So I did.</p>
The XyWrite layout is a modified XY4.KBD</code> with the letter key scancodes remapped to Colemak. The WordPerfect 5.1 layout was done by hand inside WP's keyboard editor, key by key, with the Colemak layout diagram open for reference. Download them here:</p>

XY4CO.KBD for XyWrite 4</a></li>
COLEMAK.WPK for WordPerfect 5.1</a></li>
</ul>
How to use these things</h2>
Load the XyWrite one with LOAD XY4CO.KBD</code> from the command line that you access inside the app with F5; the WordPerfect one loads from Setup → Keyboard Layout (Shift-F1 and then 5). The WordPerfect one might work with other versions besides 5.1, but why would you use any other version of WordPerfect when the platonic ideal of a word processor already exists?</p>
You're welcome, fellow weirdo. Good luck using the Pocket 8086's truly terrible keyboard... I gave up an bought a cheap external PS/2 one. It makes the whole thing a lot more fun, but it looks a little crazy spread out on a desk.</p>



Apples, Trees, and Quasimodes
2025-09-18T00:00:00+00:00
A while back, Ars Technica published a thoughtful piece about Jef Raskin</a>, tracing his long pursuit of the "humane computer" and the cul-de-sacs where that pursuit ended. It's a generous, well-told account of the designer who wanted to make machines simpler, kinder, and more aligned with the way people actually think.</p>
But part of what makes Raskin interesting is that his story isn't just Apple's story. He came out of the same cultural current John Markoff chronicled in What the Dormouse Said</em></a>—the Bay Area tradition that treated computers not as office appliances but as tools for thought, instruments of liberation. Read that way, the Canon Cat and Raskin's other projects aren't just an eccentric side quest from a frustrated Apple veteran. It's evidence of how far the humane ideal could stretch, and how quickly it ran up against the limits of commercial computing.</p>
Apple couldn't deliver Raskin's vision then, and it can't deliver it now. Neither can any other big platform company. If we want to understand why, and what Raskin still tells us about humane computing, we have to put him back in the longer lineage he belonged to, and look at how his version of the dream carried that vision but also narrowed it.</p>
Prophets and participants</h2>
What the Dormouse Said</em> documents how the Bay Area counterculture  shaped early personal computing. LSD, communes, systems theory, amorphous defense research contracts, and Engelbart's "augmentation" experiments all swirled together in a weird scene that accidentally (or maybe not</em> so accidentally) created much of the modern world.</p>
The story usually gets told with a neat list: Engelbart's demo</a>, Nelson's Xanadu</a> hypertext, Kay's Dynabook</a>, Brand's Whole Earth</a>. Xerox PARC, Steve Jobs, the World Wide Web. The familiar pantheon. But that version turns a messy, improvisational moment into a plaque. Engelbart's system needed a whole research staff just to operate; Nelson's Xanadu was (and is) more sermon than software; Kay's Dynabook lived mostly on paper; Brand mostly supplied vocabulary and vibe. What bound them together wasn't working code so much as the conviction that computers could be more than appliances and calculators, even if no one agreed on what "more" meant.</p>
Ultimately all these weird white guys had a futurist vision: computers could be liberation machines.</em> They weren't just for business automation or scientific number-crunching; they could be deployed to expand consciousness and reshape how people thought and worked.</p>
Raskin belonged to this current. Before Apple, he was an artist and a musician. He brought a humanist's suspicion of machine logic into the design lab. He argued for humane</em> interfaces: modeless, predictable, low-friction, focused on the human first. He wasn't a prophet on his own crying in the wilderness so much as another strand of the same weave.</p>
That said, his role was different than that of some of these other figures. He tried to pull those ideals out of the lab and into machines ordinary people might actually use. The Macintosh began under his hand, though what shipped was less a tool for thought than a polished derivative—what you might call a "popular religion" of computing, stripped of the harder doctrines.</p>
The Canon Cat and its predecessors were Raskin's counterargument: humane, text-first systems that tried to carry the spirit of the Dormouse</em> tradition into the commercial world without sanding off everything that made it strange. It sort of worked, but only sort of.</p>
Raskin's Humane vision</h2>
Raskin's principles are laid out most clearly in 2000's The Humane Interface</em></a>, but he'd been developing them since the late 1970s:</p>

Modelessness</strong>: eliminate modes generally, and especially when they confuse users or are hard to reason about.</li>
Quasimodes</strong>: short-lived states (like holding a key down) that don't trap the user.</li>
Humane defaults</strong>: undo everywhere, consistent commands, predictable behavior.</li>
Low cognitive load</strong>: interfaces designed around human memory and perception limits.</li>
</ul>
These ideas are recognizably part of the "Tools for Thought" tradition. Like Engelbart and the others, he wanted to reduce friction between thought and machine. Like Nelson, he believed in fluidity and extension.</p>
But there's a subtle difference. For Engelbart, augmentation meant complexity</em>: bootstrapping a system so wild it demanded co-evolution between user and tool. For Nelson, it meant endless layers of possibility</em>. For Raskin, it often meant protection</em> or constraint</em>. Humane computing wasn't only about empowerment... often it was about shielding users from mistakes, overload, and confusion.</p>
That protective impulse would shape the systems he built.</p>
Raskin's first clear articulation of his humane ideals wasn't hardware at all but The Macintosh Papers</a>, his internal proposal at Apple for a low-cost, appliance-like computer that would boot straight into a simple, modeless interface. The Mac project that followed eventually diverged—under Steve Jobs it became a graphical machine aimed at competing with the Lisa, for the reasons we've all read about</a>—but Raskin's vision was considerably more radical. He imagined a computer that behaved less like a business workstation and more like a humane, everyday tool</em>.</p>
In tone, the Macintosh Papers have more in common with Ted Nelson's Computer Lib</em> than with any corporate white paper. They read like a manifesto: plainspoken, insistent, arguing that ordinary people deserved machines that bent to them rather than the other way around. Where Nelson declared that "you can and must understand computers now," Raskin's papers laid out what such a computer should look like if you started from human needs instead of technical conventions. Both belong to that peculiar genre of the 1970s and early '80s: the computing manifesto as cultural text, half engineering and half tract.1</a></sup></p>
^{1</sup>
That genre, "photocopied computer manifesto," is very much the reason this blog exists.</p>
</div>
You could argue that the Swyft, built a few years later by his company Information Appliance Inc., was "the real</em> Macintosh" in that sense. Compact and text-first, it booted instantly, eliminated modes, and introduced the Leap keys for fluid navigation. It was Raskin's manifesto rendered in hardware. But the Swyft never made it to market; without a manufacturer to back it, Information Appliance pivoted to the SwyftCard, a fallback product that brought the same interface into the Apple II while IA waited to find a dance partner.</p>
That partner came briefly in 1987, when Canon released the Canon Cat, the only mass-produced computer to carry Raskin's humane vision into the world. The Cat retained the Swyft's defining ideas: instant boot into a blank page, consistent commands, Leap-based navigation. Marketed as a word processor, it was framed as an appliance for the office rather than an exploratory tool for thought.</p>
After its failure, Raskin returned to the same design principles in the 1990s with Archy, an unfinished software environment that tried once again to realize his humane interface on contemporary hardware. Archy never reached a finished state, but it shows how Raskin's ideas kept circling back to the same point: computing stripped down to words, presented as simply and predictably as possible.</p>
I've always had a real fondness for the Swyft/Cat lineage, and it's certainly influenced what I think a computer can be. Each one of these attempts embodied humane design: a blank screen for writing, consistent commands, no modes to trip over. The Cat in particular was radical in its way—a computer designed to feel less like a computer and more like a natural extension of the mind. It truly could have changed everything about how we use our computers had it succeeded.</p>
Unfortunately for all of us, by 1987, the market for dedicated word processors was already fading. Canon didn't seem to know what to do with the Cat—whether to sell it as an office appliance, a PC competitor, or something stranger—and the result was that it fit nowhere. Raskin's design pushed toward humane simplicity, but Canon's marketing treated it like just another machine for typing memos.</p>
It isn't surprising that it failed, though it's hard not to wonder how it might have landed a few years earlier, when the ground was more open. As it is, the Cat survives less as a commercial product than as an idea in hardware—a glimpse of what a computer could look like if the whole thing were rebuilt around text, consistency, and genuine care for the user.</p>}The Paradox of Openness</h2>
The Cat also embodies why Raskin's philosophy was not necessarily on the same wavelength as some of those other visionary systems. On the surface, the Canon Cat looked open. It booted to a blank screen. Everything was text. You could jump anywhere, edit fluidly, undo anything. Compared to the modal labyrinth of DOS or early Mac software, it felt like freedom.</p>
But look closer and you see the narrowing. The Cat gave you fewer ways to improvise. Its humane design was also constraining</em> design. It reduced your options in order to keep you safe.</p>
The real irony is that the Cat wasn't even truly closed</em> in the way a smartphone or Chromebook might be considered so today. Underneath, it ran on a Forth environment. You could, if you knew how, drop into Forth and even program directly in 68k assembler. In principle, it was as open as any hacker could want, at least from a software perspective.</p>
The catch was cultural, not technical. From the Ars piece:</p>

IAI's back door to Forth quietly shipped in every Cat, and the clue was a curious omission in the online help: USE FRONT-ANSWER. This otherwise unexplained and unused key combination was the gateway. If you entered the string Enable Forth Language</code>, highlighted it, and evaluated it with USE FRONT-ANSWER (not CALC; usually Control-Backspace in MAME), you'd get a Forth ok</code> prompt, and the system was now yours. Reset the Cat or type re</code> to return to the editor.</p>
</blockquote>
Canon didn't provide documentation that would have made that power accessible, and Raskin's design philosophy treated it as outside the normal use case. Extensibility was there if you knew where to look for it, but it wasn't encouraged. The humane interface was meant to keep most users away from the hood, even though what was under the hood was remarkably open.</p>
That makes the Cat's paradox sharper: it was a genuinely extensible software environment (up to a point) presented as a sealed appliance. The hardware mostly was</em> a sealed appliance. Contrast this with Emacs or Smalltalk, where openness is the posture of the environment itself. You are expected to extend and reshape as you go, building your tools out of themselves. The Cat offered the same possibility--Forth is a remarkably flexible language, especially for microcomputers--but it discouraged you from taking it.</p>
Humane computing, in Raskin's hands, edged toward hermetic</em> computing. He built openness in, but sealed it away behind an interface designed to keep it out of sight.</p>
Cul-de-sacs vs. Branches</h2>
All of this, to me, is why calling Raskin's systems thinking a "cul-de-sac" misses the point, and is the wrong way to think about his legacy.</p>
If "cul-de-sac" means "product that didn't sell," then sure, the Cat and SwyftCard qualify. They were total dead-ends. But by that same measure, Engelbart's NLS, Nelson's Xanadu, Kay's Smalltalk, or even Lotus Agenda</a> are dead-ends, too. By that measure, most of the "Tools for Thought" tradition didn't lead anywhere.</p>
The reality is different. These systems were branches</strong>. They were rhizomes, in the Deleuze and Guattari</a> sense. They didn't reach the mainstream, but they seeded ideas that echoed elsewhere, connecting threads that run throughout the history of computing. Hypertext, graphical interfaces, undo, modeless editing—all of these survived in one form or another.</p>
Raskin's branch is no exception. His machines exposed a fundamental tension inside the tradition: how far do you go in protecting the user from complexity? At what point does "humane" become "hermetic"? Those questions didn't vanish with the Cat. They're still with us every time a productivity app promises "simplicity" at the cost of agency.</p>
Raskin's humane ideals live on in obvious ways, to the benefit of anyone using a graphical computer today—undo everywhere, discoverability, and consistent commands and shortcuts are now interface common sense. But the deeper thread, the ethos that inspired him and others in the tradition of computers as tools for thought, survived mostly outside the mainstream. It persists in systems that never had to sell millions of units or satisfy quarterly targets, that never had to justify their existence to the mass of people using PCs—tools that could afford to remain strange, open, and humane on their own terms. Emacs, Oberon, and Smalltalk belong here, but so do newer experiments like Uxn</a> and 9front</a>.</p>
The Cat failed partly because it tried to straddle two worlds: commercial appliance and humane machine, whereas something like Emacs survives precisely because it never had to. It's as complex as you want it to be.</p>
This is the sharper point: radical, humane, exploratory computing never survives in the mainstream. The mainstream is built for profit and predictability. Even Engelbart's work was DARPA-funded, not venture-backed. When you put humane ideals through commercial constraints, they collapse into simplistic appliances, the "For Dummies</a>" version of the original intent. That doesn't mean the tradition is dead. But it does mean you have to look off to the side, away from the market's center, to see it alive.</p>
The dilemma(s)</h2>
Raskin's story sharpens two dilemmas that haven't gone away.</p>
The first is practical: make a system too open, and it risks being overwhelming. Make it too humane, and it risks narrowing into something sealed and hermetic, and not useful enough. The Cat, while also a victim of other factors, tried to balance the two and ended up fitting nowhere.</p>
The lesson isn't that humane computing is impossible. It's that humane computing can't just mean protective</em> computing. It has to mean trusting users with both simplicity and openness. That's why Org mode and even Mac System 7</a> endure and the Cat does not.</p>
The deeper implication is harder, but maybe truer: the true Tools for Thought we still wish existed will never come from Apple, Microsoft, Google, OpenAI, or any other large player in the software or hardware space. They can't</em>. These companies' scale and incentives point elsewhere—toward lock-in, surveillance, and products that are safe enough to sell but never open enough to empower. The logic of scale makes them constitutionally incapable of building systems that are truly humane and open. The next humane systems, if they arrive, will have to come from outside those walls, as they always have: from margins, from hobbyists, from research labs, and from stubborn communities of practice. But as those platform companies make it more and more difficult to experiment, how do we keep pushing these philosophies forward?</p>
Jef Raskin's philosophy isn't a cul-de-sac in computing history. He's responsible for a branch of the "Tools for Thought" tradition—a branch that shows both the promise and the peril of humane design. His machines make clear how far you can go when you put the human first, and how easily that ideal can collapse into constraint once it's pushed through commercial channels and turned into walled gardens.</p>
The humane thread survives, but only outside the center—in the tools that don't have to answer to quarterly earnings, in projects that refuse to die just because they don't fit the market. The Dormouse</em> lineage isn't gone. It just doesn't live where the money is, because it can't. If you want your computer to be humane in the deeper sense—not an appliance, but an instrument for thought—you have to look to the margins. That's where it has always been, and where it still is today. If it survives, that's where it'll still be.</p>


Hidden Agendas, Lost Cities
2025-08-14T00:00:00+00:00
I first really spent time with Lotus Agenda was because of a guest post I wrote for 512 Pixels</a>, which meant I approached it with the half-serious curiosity of someone who'd long harbored a suspicion that MS-DOS software might change the way he thought.</p>
I'd been Agenda-curious for a long time, but never tried to actually use it for real. I booted it up on my Pocket 8086, watched cards crawl into place, watched queries resolve themselves into slices of sense, and realized with annoying clarity: none of the modern "task management" apps I've used---not Todoist, not OmniFocus, Amazing Marvin, Remember the Milk, not Microsoft To Do---could do any</em> of this. Not really. Agenda wasn't "a to-do list." It was a cognitive environment</em>, a place where items could belong to more than one thing at once, where priorities didn't march in a line but emerged</em> from the way pieces related to one another.</p>
This is not a user guide (Tavis Ormandy already wrote a careful, generous one here</a>). It's more of a lament---or maybe more of a small civic memorial---for an unexplored avenue in productivity software that modern tools only approximate in fragments. The utopian tools-for-thought vision of Agenda didn't vanish because it was wrong; it vanished because it was weird</em> and unprofitable</em>.</p>
The city that could rearrange itself</h2>
Agenda arrived in the late 1980s, back when DOS reigned, mice were a nice-to-have, and "personal information manager" still meant something closer to augment your mind</em> than make a dashboard</em>. Part of the program's strangeness is conceptual: it treats pieces of information as first-class citizens, and treats "organization" as a set of overlapping relationships you can keep recombining on the fly. You don't file</em> a thought, you assign</em> it---often in many places at once---and then you ask the system to show you different views of the mess based on rules you declare after the fact. (The rules are the interesting part; the rules are where the thinking is.)</p>
If you want the canonical period feel, the original manuals live on the Internet Archive---start with the Users Guide</a> and leaf through "Quick Start," "Working with Macros," and "Working with Definition Files." It's not just software; it's a stance</em> toward information, one where you retain permission to change your mind about how things fit together long after you've written them down. (In a modern app, that's called "refactoring a database schema." In Agenda it's called "Tuesday.")1</a></sup></p>
^{1</sup>
This "it's just a weird database" structure may be familiar to you from another, much more widely maligned, Lotus application: Notes. I used it at a long-ago job, and it was far past its prime by then, but one of its strengths was also visible in Agenda: Lotus Notes was really an application creation environment where everything in your Notes database was really just part of a huge no-SQL store of stuff</em>. It made it a very strange email client in the Gmail/Exchange world we live in, but before those days, it was extremely powerful and lots of big enterprises used it to build their own custom knowledge tools.</p>
</div>}The old community documentation still reads like a transmission from an alternate reality. Bob Newell hosts a trove of material, including the very plainspoken Agenda FAQ</a> and an index of resources</a>. The model boils down to three ideas: items</strong> (free-form text), categories</strong> (overlapping conceptual buckets), and views</strong> (query-driven windows that splice those relationships into meaning). It sounds mundane until you realize what it enables: you can tag promiscuously, look laterally, and conjure new slices of your work without</em> reorganizing the work itself. The structure is a lens, not a prison. You can do whatever you want with the stuff you have in Agenda, and once you see that, you can't really un-see it.</p>
If you need a quick fact dump beneath the romance, the Wikipedia entry</a> puts names and dates to things, and you can always go download the thing at WinWorld</a> and look at screenshots. But the encyclopedia tone, and even the manuals themselves, can't convey the feeling of it: Agenda behaves like a little city that rearranges itself when you ask a new question.</p>
GTD, capture, and the tyranny of places</h2>
When Getting Things Done</em> landed in the early 2000s, it didn't just become a bestselling book---it became a lifestyle, especially in Mac and designer circles. The method spread through blogs, podcasts, and the burgeoning world of productivity forums. Merlin Mann was explaining it on 43 Folders, OmniGroup was building OmniFocus as the "Mac-ass Mac app" for it (full disclosure: I was one of the early users on the invite-only "Sneaky Peek" builds of OmniFocus 1, so I know whereof I speak), and every independent developer with a copy of Xcode seemed to be designing their own GTD-inspired task manager.</p>
The appeal was obvious: GTD promised to tame life's chaos with a deceptively simple premise: capture absolutely everything</em> that has your attention, process it into a trusted system, and review it regularly. In the design community, where aesthetics and workflows intertwined, GTD became more than a system; it was the design brief for a whole generation of task apps. It was also the progenitor of a million blogs and podcasts, "productivity hacks," and a bunch of other things, including the widely-misunderstood but brilliant Inbox Zero</a>. Interfaces were built around inboxes, projects, and contexts because that's what the method demanded. Soon, the GTD-shaped template was the default for "serious" productivity software.</p>
But baked into GTD's promise is a subtle, rigid assumption: if you're going to have everything</em> in your system, that system has to have a place for everything. Which means you need to decide---often very early</em>---what those "places" are. And then you have to keep</em> deciding.</p>
Once you have that structure, it starts shaping the way you think about the stuff you put into it. Capture leads to categorization; categorization leads to a fixed set of buckets; the buckets become the mental map. It's clean, and it works---if you're comfortable letting your first pass at organization become the scaffolding for everything that follows.</p>
The obvious counterpoint is that GTD is supposed to be as flexible as you want it to be. You can rename your contexts, refactor your projects, blow away your tag structure entirely. Nothing is technically fixed. But in practice, that flexibility invites endless "tuning"---fiddling with categories, reorganizing lists, tweaking perspectives---all of which feels</em> like productivity but often isn't. (Merlin Mann, years after popularizing GTD, would openly joke about this "meta-work" trap: the point where the system stops serving you and starts existing just to be fussed with.)</p>
Agenda never asked that of you. It didn't care how you captured things, or whether you processed them at all. You could dump in whatever you had---fragments, ideas, tasks, raw notes---and worry about what they meant later. The categories, the queries, the "workflow" could emerge</em> from the material itself. That's the part that doesn't survive in GTD-land: the permission to let meaning arrive late, to let your structure be provisional until the content tells you otherwise.</p>
Modern GTD-inspired tools enforce the place-for-everything worldview because they have to: it's the only way to make capture-and-process work at scale. Agenda never gave a damn. And that indifference to workflow---that refusal to tell you what your process should be---is exactly what made it so weird, and so powerful. (Also so hard to learn, as someone whose brain has been poisoned by about twenty years of GTD knowledge at this point.)</p>
"But what about Org mode?" (A fair question with a short answer)</h2>
Org mode belongs in this story, but not as the hero---more like a secret passage through the ruins. Its agenda views, tags and properties, and query-driven lists absolutely echo Agenda's model; if you squint, it's the same "write first, declare meaning later" ethic, implemented in plain text with Emacs as a friendly (well, for certain definitions of "friend" anyway) exoskeleton. If you already live in Emacs, Org's fluency with emergent structure feels like a private privilege in a world of tidy linear apps. If you don't, it feels like an alien monolith.</p>
But Org's virtues are also why it can't carry the argument here. It's a niche</em> on purpose. It asks you to believe in it. It's not trying to be a mass-market replacement for flattened software; it's trying to make your thought-work livable on your terms. (Which, ironically, is the real shadow of Agenda: not a product, but a posture.</em>) The lament isn't that Org isn't popular enough; the lament is that the mainstream either forgot this family of possibilities existed at all</em> or, more likely, wouldn't have cared even if they'd been exposed to it.</p>
If you want one modern artifact that shows the old spirit survives in the open, Tavis's post is it---a living museum of the mechanics and the ideas without the romantic haze. But even there, you can feel the distance between "here's how it works" and "here's why the world won't pay for it." Agenda's lineage is just a subculture now, and subcultures are where weird things thrive quietly</em> until the zeitgeist tilts (if it ever does).</p>
What we lost (and why it was worth talking about)</h2>
Agenda teaches two lessons that most modern tools actively discourage.</p>
1) Write first, decide what it means later.</strong>

In a GTD-shaped app, the categories define the work. In Agenda, the work could define the categories. You could over-assign items to overlapping categories, then build views that treat the same pile of stuff as a to-do list, a knowledge base, and a status report---without rewriting a thing. The "truth" of the item was relational, not essential. It tolerated ambiguity and rewarded curiosity, letting meaning emerge from the material instead of forcing you to decide too early.</p>
2) Use queries as lenses, not cages.</strong>

Agenda's views are invitations to ask new questions of the same corpus, not fences you're forced to paint every week. If you've used any modern PKM system with a decent query language you've tasted this, but in Agenda it's foundational: the query is</em> the interface. Most of us now expect a list with a filter. Agenda expects a question with an answer that can keep changing. The difference is small until it isn't.2</a></sup></p>
^{2</sup>
The original 1.0 OmniFocus perspectives were a work of art. They weren’t just filters, they were saved states of your working mind---complex, multi-layered lenses that could flip your whole world from "what can I do in the next 5 minutes at this coffee shop" to "what am I avoiding that I said I cared about last week." They rewarded tinkering, but they also rewarded trust; you could live in them without having to think about the plumbing underneath. Naturally, they were replaced by something dumber in OmniFocus 2.0 and have never returned. I've never really found any single task-manager view I liked as much as my beloved main OF perspective.</p>
</div>
Why does this matter? Because software shapes thought, and mainstream tools teach us that "productivity" means "decide a structure early and obey it." (Which is fine for keeping up with logistics, but it's deadly for discovery.) Tools in the Agenda lineage teach a competing lesson: that you can improve your thinking by letting your structures arrive late</em>, by learning from how the work wants to group itself, by refusing to declare premature victory over meaning... and doing so without having to manage a rat's nest of hyperlinks like you do in a tool like Obsidian or Logseq. Rheingold's Tools for Thought</em> history of mind-expanding systems glows with this ethos---computing not as a factory but as a prosthesis for the weird parts of people's intelligence. Agenda was one of the last mainstream artifacts that tried to embody it in everyday work.</p>}The little memorial at the end</h2>
While I was trying to use Agenda, I'd shut down the Pocket 8086 (or DOSBox-X) and open a modern app and would feel (for a second) like I'd stepped from a city that could rearrange itself into a grid of cul-de-sacs, like I was returning from Tlon, Uqbar, and Orbius Tertius</a>. The modern apps are fine---they keep you busy, get you home on time, keep you from forgetting to water the houseplants on the days you're supposed to do that---but they rarely let your work teach you how it wants to be seen. That's the loss. There's simply no widely-available tool that captures the same spirit, and it seems unlikely that there ever will be another one. Everything is a checklist now, because everything is a checklist now.</p>
Agenda's utopia didn't die because it was wrong; it died because it was weird.</em> And yet, in the corners (an Emacs buffer here, a vintage manual there, an enthusiast's blog that reads like field notes, the long-abandoned open source attempt to recreate it</a>), it's still alive enough to haunt the imagination. If you've ever suspected that the best part of "productivity" is the part that changes your mind</em>, Agenda is waiting---not for you to adopt it permanently, but for you to feel what it's like to live in a tool that expects your structures to arrive late. Problem is, you need a DOS machine to explore it.</p>
It's a small thing, a private thing. But it's the difference between software that manages actions and software that amplifies thought</em>. And once you've seen how it could be, you'll always miss it.</p>
If you want to explore the ruins</h2>
A few places to wander, if you're in the mood:</p>

Tavis Ormandy, "Lotus Agenda."</strong></a> A modern, careful tour with working examples and lots of "oh that's</em> how it fits together" moments.</li>
Lotus Agenda on Wikipedia.</strong></a> Names, dates, versions---the factual scaffold beneath the romance.</li>
Original manuals on Internet Archive.</strong></a> The best way to feel how function-key-forward, query-first, and gloriously un-app-store the whole thing is.</li>
Bob Newell's Agenda resources</a> and FAQ</a>.</strong> Dry as saltine crackers on the set of Ishtar</em></a>, but they give you the mental model in plain language.</li>
Howard Rheingold, Tools for Thought</em>.</strong></a> Readable online; a window into the ethos Agenda came from.</li>
</ul>


A New Old Idea
2022-12-24T00:00:00+00:00
I finally decided to sit down and collect my thoughts about permacomputing, and the main thing I discovered was that I need to do more thinking.</p>

Contents</h3>

I. The seed is planted</a></li>
II. The Dream</a></li>
III. The hardware</a></li>
IV. The software</a></li>
V. What's next?</a></li>
</ul>
</figure>
The seed is planted</h2>
I've been thinking about permacomputing</a> since I first read Steve Lord</a>'s excellent piece about "heirloom computing</a>" quite some time ago. In all honesty, it's something I've been thinking about since I was in high school in one way or another, when I would print out old Altair and Fairchild F8</a> manuals (because I thought I might try to write a Channel F</a> game, of course) and read them obsessively.</p>
Ever since I first saw a picture of the Information Appliance Swyft</a>, which eventually became the Canon Cat</a>, I've had in my head that something like that</em> is what I should be using instead of the laptops and desktops and software stacks I've been using all along. My formative computing memories are watching my mainframe-programmer grandfather show me how he wrote BASIC programs on his Packard Bell PC with an amber monochrome monitor, playing Oregon Trail</em> and Number Munchers</em> on Apple //e computers (with color monitors!) in my elementary school's computer lab, using ClarisWorks to create "books" and newspapers and write.</p>

    
    
        Figure 1: An Information Appliance Swyft prototype.

        Image from Digibarn</a>.
    </figcaption>
</figure>
I took a class in third grade where they taught us how to use Yahooligans!</a> to find things on the Internet, and I played with HyperCard</a> whenever I could. Once we had the internet at home, I used it to learn about computers that had already passed into obsolescence---Intellivisions and Ataris, Altos and Novas, Crays and Thinking Machines. By eighth or ninth grade I was collecting old computers however I could get my hands on them, especially 680x0-based Macs. I fell in love with the carrying handles on the old compact Macs, and with what they implied about what a computer was supposed to be used for.</p>
Some of what captured my attention about all these old machines was nostalgia, for sure. But more than that, it was a sense that there had been a fork somewhere. What drew me to old computers, and what still</em> draws me to them after all these years, is the sense that people who used them new had a different faith in them: of computers as a tool for thought</a>, as something that people could use to accomplish something new. The idea of computers as "the bicycle for the mind" felt like a lost future, the same way ring-shaped space stations and supersonic airliners were promised and lost.</p>
From there---the feeling of a limitless future, of the promise of the computer as a thing to make people more themselves</em> and to somehow enable their users to live a fuller life, led me to computer history books like Dealers of Lightning</em></a> and The Soul of a New Machine</em></a>, to people like Alan Kay, J.C.R. Licklider, Ted Nelson, to the Memex and the NLS and to Smalltalk, to Lisp and Scheme and Emacs.</p>
But I never forgot the way I felt clicking through pictures of the Swyft and reading about how its users could drop down into its Forth environment or even to 68k assembly language programming at will.</p>
The Dream</h2>
I was down a rabbithole reading about CP/M-86 (long story) when I came up for air and posted this on Mastodon (yes, with a content warning, which marks the image as "sensitive content" even though it's a photo of a computer, because Mastodon is Mastodon):</p>

    </iframe>