A while back I had an interesting conversation with my older son, who is in 7th grade. He was telling me about various new rules his school had introduced — like only two bathroom breaks per week per class — which, we agreed, did not make much sense. But then he added: It seems like the teachers also know that the rules don’t make sense. If you talk to them about it, it’s pretty clear that this is something that they’ve just been told they have to say. Then we all forget about it and they go back to teaching. It’s not like anyone is checking if they actually enforce it.

I’ve thought about this conversation now and then in recent months. It seems to encapsulate, in a small way, the professional autonomy that has somehow become one of the central political battlegrounds of our time.

Teaching is very hard to manage from the top down. As a teacher myself, I’ve often experienced this. There are all kinds of rules and standards that are announced from the top. But very little of what you do in the classroom can be effectively monitored. In reality, you teach your classes according to your own standards, and follow the rules that make sense to you.

This, I think, is why teaching is the quintessential public service. Which is something different from a public good.

Education is not well suited for market provision, for reasons that are probably obvious to most people reading this. It doesn’t produce a distinct commodity, that can be owned and exchanged. The product, such as it is, is almost impossible for the “consumer” to meaningfully evaluate. How do you know what you need to learn, before you’ve learned it? And of course there are externalities, economists’ favorite argument for public provision. The benefits from an educated population are broadly shared.

But the problem is not just that the product of education does not look like a commodity. The process is also a problem.  Even if we think of teaching as just another form of production, it’s very difficult to rationalize it in the way that other kinds of work can be. You can’t standardize the inputs and conditions of production, which is the key to successful automation. Teachers have to make all kinds of decisions, on the spot, in unpredictable conditions. (Does this kid really have to go to the bathroom?) And there’s no straightforward way to say which decision is the right one. You have to rely on teachers to exercise their own judgement, and evaluating outcomes on the merits. Which means, fundamentally, relying on intrinsic motivation rather than external direction or uniform rules.

I’ve been teaching college for a dozen years now. Anyone who’s done this work knows how rewarding it can be when it goes well, and how agonizing it can be when it goes badly, and how hard you will work to do better. But none of these outcomes can be measured  or enforced by a boss. Assessments are a joke, we all know that. The nature of the work is that the best you can do is make sure that teachers are motivated and have the resources they need — and, yes, get rid of the really bad ones — and then get out of their way.

This is not only true of teaching, though teaching is certainly among the largest and most visible forms of work that depend so strongly on the autonomy and intrinsic motivation of the worker. This characteristic, it seems to me, is a central, though seldom articulated, reason for public provision of all kinds.

I am, professionally, an economist. We economists, and economist-influenced policy people, are used to talking about public goods. We have a clear language for that. We are less used to talking about public provision.

It’s one thing to argue that government should ensure that everyone has access to education, or health care, or childcare. It’s a different, distinct argument to say that these things should be performed directly by public employees. This second argument hinges on the need for autonomous, intrinsically motivated decisions by the people doing the work. Intrinsic motivation is the opposite of incentives, indeed it requires insulation from them. It requires a space where the person doing the job can freely make decisions based on their own professional judgement — the space that things like civil service protections are precisely intended to preserve.

These questions are not limited to the public sector. They are the most politically salient there, at the moment, since the Trump-Musk regime is practically defined by attacks on the civil service.¹ But the autonomy of workers within the production process exists in all kinds of settings, public and private; complete deskilling and perfect supervision are never possible. And this defines an axis of conflict, or a dimension of socialization, which is largely orthogonal to the question of public versus private ownership.

Even for-profit corporations depend on intrinsic motivation – people’ desire to do their jobs – and the recognition of the manager’s authority as legitimate. This is true wherever ongoing coordinated activity is required.

As a productive community and a polis, the corporation depends on what David Graeber calls “baseline communism”: the principle that when one member of a community needs or requests something within a normal range, another member will provide it without the need for any explicit reward or punishment. Within a workplace – even a rigidly profit-oriented one – tasks often must be performed by whoever is in a position to, and tools provided to whoever needs them. When one office worker asks another to use their stapler, the answer will not be “what’s in it for me?”. The guiding principle here is, from each according to their abilities, to each according to their needs.

A distinct but related category of intrinsic motivation is what the French historian Jules Michelet called “the professional conscience” – the disinterested desire to do one’s job well. Studies of people’s experience of the workplace reveal plenty of alienation and insubordination, but also a great deal of effort to carry out the work – whatever it may be – as well as possible, by its own standards.

Some degree of worker autonomy is always necessary for the routine functioning of production. But how much autonomy is always a site of conflict, often latent, occasionally acute.

This is a point emphasized by observant historians of socialism. George Eley, in his monumental history of the left in Europe Forging Democracy, emphasizes that in the great revolutionary upsurge that immediately followed the First World War, socialization meant something quite different from public ownership. The central goal of the movements that came to be known as “council communism,” in Germany and Italy in particular, was establishing workers’ control over the production process, regardless of formal ownership.  Socialization in this context meant a change in the internal organization of the workplace, rather than a change in who exercised authority at the top. In Eley’s words:

The distinctiveness of revolutionary activity in 1917–23 lay in the workers’ councils… These ranged from unofficial strike committees developing larger political aims, like the shop stewards’ movements of Clydeside, Sheffield, or Berlin, to sophisticated revolutionary innovation, like the factory councils in Turin. In between came a rich assortment: the Ra ̈te in Germany and Austria, claiming functions of class representation in a locality; councils based in factories, firms, or other economic units; and local action committees for specific ends…

A new medium of working-class activity, councils differed from both socialist parties, which acted through parliamentary and state institutions, and unions, which worked on the capitalist economy’s given assumptions via the wage relation. … Stronger versions of the council idea were hostile to orthodox trade unionism and socialist electoralism, recoiling from the accepted model of separately organized, centralized, nationally focused political and economic movements. Instead, councils were based within production: inside the unit of production itself, in the factory, the plant, or the shop. Councils raised issues of industrial democracy, workers’ self-management, and workers’ control.

Unlike unions, councils were not imagined as vehicles for collective negotiation with the boss over workers’ specific interests as workers — pay, working conditions, and so on. Rather, they were imagined as vehicles for replacing the boss and organizing the production process itself.

Council communism was a product of the breakdown of established hierarchies that followed the war; as a distinct movement, it was short-lived. But the axis of conflict it crystallized — over control of the production process, as opposed to workers’ interests as sellers of labor power — is very much still with us. And as teachers know as well as anyone, workplace autonomy needs to be defended from the state as well as from private employers.

*

One wouldn’t want to say there is no relationship between the two sense of socialization. The pursuit of profit creates powerful pressures for the erosion of workers’ control. While public ownership is not the same as workers’ control over production, it can be a shield for it where it already exists. As a CUNY professor, I am well aware of the benefits of an insulating layer of bureaucracy. Those of us in the classroom work very hard to produce learning in our students. People at the top of the institution have all kinds of ideas about how to change things, generally for the worse; but they usually bog down in thick layers of bureaucratic inertia before they can do much harm.

The social position of teachers and other professional employees has taken on new urgency since the election. One reason for this is that they are in the crosshairs of the right. But another reason is that the left is not sure how it feels about them, either.

A frequent topic in election post-mortems is the disproportionate share of votes the Democrats now get from the college-educated. In a typical exit poll, the split was almost exactly symmetrical: 56-42 for Harris among those with college degrees, 56-43 for Trump among those without.²  It’s a commonplace in these discussions, perhaps especially on the left, to contrast college-educated with working-class, as mutually exclusive categories.

Here for example is a Jacobin piece that treats the two categorizations as straightforwardly equivalent: “Donald Trump made substantial inroads among the working class in November. The best data currently available from AP VoteCast indicates that the Democrats’ share of non-college-educated voters fell from an already low 47% in 2020 to 43% in 2024. …. Democrats lost among working-class (noncollege) voters.”

As a factual matter, it is clearly true that Democratic voters in the US are increasingly found among those with higher education. Nor is this a phenomenon unique to the US. Thomas Piketty highlights it as a general phenomenon in his widely-quoted formulation of the “Brahmin left.”

But who exactly are the college-educated?

One way to answer this is to look at the US occupations with the greatest difference in employment as a share of those with college degrees, and those without. Count up those with degrees, subtract those without them, and you have your Brahmin occupations. If you do this exercise, number one on the list turns out to be nurses; number two is elementary school teachers, with secondary and college teachers a bit further down. If you add preschool teachers (0.5%), special-education (0.7%) and “other” teachers (0.7%) to the elementary and secondary teachers, about 10% of all American workers with college degrees are classroom instructors of some kind.

As a first approximation, then, when you talk about “college-educated voters,” who you are talking about is nurses and teachers. And if, like the person in Jacobin, you write “working-class (noncollege) voters,” what you are saying is that teachers (and nurses) are not members of the working class. Which presumably means that they are not workers.

Well, are teachers workers? My impression is that there is some uncertainty about this. And if not workers, then what?

The obvious answer is: Teachers are members of the professional-managerial class (PMC). Indeed, when Barbara and John Ehrenreich coined this term back in the 1970s, they offered teachers as the paradigm case.³

It’s interesting reading the Ehrenreichs’ essay today. Its starting point is the fact that “in the United States in the last two decades, the left has been concentrated most heavily among people who feel themselves to be ‘middle class’.” At that time, “the last two decades” would begin in the 1950s. The Brahmin left is evidently not a new development.

For the Ehrenreichs, PMC members like teachers are distinct from the working class because their work is about reproducing the existing social relations, rather than producing particular commodities. If we want to place someone in one or the other of these categories, we should ask: “Is their function required by the process of material production as such, or by capital’s concern for ruling and controlling the productive process?” On the face of it, this seems clear enough. But with a little more thought, we might wonder  is this a distinction between teachers and nurses as opposed to real workers? Or does it rather reflect a dimension of all work under capitalism?

Not all off our useful and necessary activity can be embodied in discrete physical objects with clear property rights attached to them. And the maintenance of capitalist relations of domination and control are, it seems to me, fundamental to all kinds of work, not just that of a specialized group. We also might want to consider the distinction — elided in the Ehrenreich essay — between reproducing capitalist social relations specifically, and social reproduction in general. Surely the specialists in the former must include the police, and guard labor generally — but cops are seldom if ever who people have in mind when they talk about the PMC.

Consider the opposite end of the scale of occupations above — those that account for a disproportionately large share of those who don’t have college degrees, relative to those who do. Number one by this criterion is truck drivers, accounting for 3.6 percent of workers without a college degree and just half a percent of those without. Just behind them are cashiers, at 3.5 and 0.5.

Driving seems straightforwardly to be material production: a commodity located here is different from a commodity located there. But what about cashiers? They come into the story only once the process of material production is done with; as far as that is concerned, the users of commodities could simply claim them directly, if necessary leaving some record themselves. (You can still occasionally find this arrangement in small New England country stores.)

The cashier is needed not for production, but to ensure that the use-value is exchanged for an equivalent quantity of money. Surely, by the Ehrenreichs’ criteria, this is a perfect example of work that exists to maintain capitalist social relations, rather than anything to do with the needs of material production? But nobody, when they talk about the PMC, is thinking of cashiers.

The Ehrenreichs, in the 1974 essay, do offer a second criteria for membership in the PMC: a class, they write, is also defined by :a coherent social and cultural existence: members of a class share a common life style, educational background, kinship networks, consumption patterns, work habits, beliefs.” The problem, which they acknowledge but don’t really engage with, is that classes defined in this way don’t coincide with those defined by one’s role in production. It may perhaps be true (or perhaps not) that the spouses and parents of teachers are likely to be professionals rather than blue-collar workers; it may be true (though one shouldn’t take it for granted) that teachers have more in  common culturally with property owners than with people who use their hands for a living.

My own experience teaching CUNY students suggests, for what it’s worth, that the Ehrenreichs were wrong on this point. My students, almost all non-white immigrants whose day jobs are as  busboys and doormen and home health aides and taxi drivers, are immersed in the exact same culture as my own kids are. Culturally there is no difference my students and middle-class kids; the only difference is my students’ lack of economic security. To transpose the well-known quote: the poor are no different from you or me, except they have less money.⁴

That is not how it looked to the Ehrenreichs. For them, the conflict between PMC and workers is the same kind of fundamental opposition as between workers and capitalists. “The relationship between the working class and the PMC, they write, “is objectively antagonistic.” To Marx and Engels’ classic list of historic oppositions — “freeman and slave, patrician and plebeian, lord and serf” the Ehrenreichs  add “teacher and student, …  social worker and client.”

As a teacher, I must admit that I am not sure that my relationship with my students is useful compared with that of a feudal lord to his serfs. (But then I wouldn’t think so, would I?) Again, one might note that the Ehrenrecichs — like later users of the PMC concept — don’t  distinguish between the reproduction of specifically capitalist relations, and social reproduction in general. As a parent, am I producing labor power for the capitalist class? Well, yes. But I don’t think that is all I am doing.

Be that as it may, there do seem to be  people who do seem to see professionals like teachers as the class enemy. Hostility to the PMC is particularly central to the left-right hybrid politics expressed in places like Compact, or American Affairs or in different forms by Vivek Chibber on the left and J. D. Vance on the right — what people sometimes call diagonalism.

We might also call it Wagenknechtianism, after its most distinctive practitioner in European politics. Sahra Wagenknecht herself justifies her eponymous party on the ground that she is the tribune of “the little people, those in small towns and villages, without university degrees, …  the world beyond professional political life,” as against the “new, university-educated, professional class.”

One can’t disagree that there is a problem when political officials and activists become a small, self-contained group, when politics becomes simply a profession. (Though it’s a bit funny coming from someone like Wagenknecht, who has done nothing else in her life.) But this, it seems to me, is a different issue from the share of voters drawn from the college-educated. A professional politician is one thing; an elementary school teacher is something else.

Certainly teachers occupy an ambiguous role in contemporary capitalism. The PMC is one way of theorizing that. But one could also think of teachers as somewhat analogous to factory workers in semi-industrialized countries. Their strategic position does, objectively, make them relatively privileged compared with the majority of the population. But it also gives them a basis of social power. Both factory workers and teachers provide a useful service for the bosses. (That’s what the money is for.) But their distinctive work experiences can also build solidarity, embody anti-capitalist values, and prefigures alternative mode of social organization. This is, perhaps, as true of the work of teaching as it is of work in factories.

Somewhere in his prison writings, Antonio Gramsci describes a conversation with some Sardinian soldiers who were brought to Turin to help put down the great strikes of 1921. “What have you come to Turin for, he asked them.  “We’ve come to shoot at some gentlemen who are going on strike.”

“But it’s not the gentleman who’re going on strike, it’s the workers, they’re poor people.”

“Here they’re all gentlemen: they wear collars and ties and earn 30 lire a day.”

Present-day professionals wear collars and ties; we make the contemporary equivalent of 30 lire a day. It’s hard not to be reminded of Jay Gould’s perhaps apocryphal claim that “I can hire one half of the working class to kill the other half.” In this context, Wagenknechtian talk about the professional class sounds like a job application.

*

People who talk about the PMC tend to be somewhere in the Marxist tradition. If we look to Marx as a political strategist rather than a class taxonomist, then his great insight was the need to link a positive program to some objective force able to advance it. Politics, from this point of view, is about giving conscious, organized form to the conflicts that already exist. Applying this insight today means recognizing that the lines of conflict are different than they were in Marx’s day.

Professional conscience is an important source of power for left. Our side cannot organize on basis of money. Money as an organizing principle works for a program of advancing or stabilizing the power of the bosses; it doesn’t work for a program of challenging that power. The power and prestige of technical expertise are, in principle, more amenable to a program of social transformation. The desire to do one’s job well is something that capitalism cannot do without. And that creates an alternative basis of solidarity and social power — the possibility of what Veblen long ago imagined as a “Soviet of engineers”, mobilized against the “sabotage” of production by the owning class.

A soviet of engineers may not sound very plausible at this moment, to you or me. But it’s worth recalling that to the lords of Silicon Valley, it seems very plausible. In a conversation with tech reporter Kara Swisher, Ezra Klein suggests that Musks role in the government, and the broader tech turn to the right, stems from the fact that “a lot of the C.E.O.s just hated their employees. And what radicalized them was that they had lost control of their companies, and they wanted that control back.”

You can say that CEOs had never lost control of their companies. You can say this claim, or Marc Andreessen’s even wilder claim that many tech companies “were hours away from full-blown violent riots … by their own employees” sounds paranoid and hysterical.

On one level, they are. But they also express something real. Professional employees — teachers, engineers, coders — necessarily have a degree of autonomy on the job, a space within which they decide what needs to be done. To that extent, the owners, when they need to make use of such employees, do indeed lose control of their companies. Professional norms, standards, credentials, skills — these are real sources of power for those of us who do the work, as against those who claim the results of it.

Historically it has always been relatively privileged workers who lead the opposition to the bosses. That’s who’s in a position to do it.

It is true that, today, voters for traditional parties of the left are disproportionately likely to be college-educated. But this is also increasingly true of union members. In the United States today, 47% of union members have college degrees, as opposed to 41% of the population as a whole. In the US today, there are one million union members in manufacturing. There are three million in education. You can say that’s a problem. You can also say, it’s a base we can build on.

Elon Musk and his peers hate their workers. They hate what they see as their unjustified power over production; unjustified, from their point of view, because it is not based on ownership. Whether it is based on skills or credentials or regulations or union membership doesn’t matter to them; in some contexts, arguably, it shouldn’t matter so much to us.

The overlap between professional workers like teachers and union members isn’t just an abstraction; in the past decade, a disproportionate number of strikes have been carried out by teachers. These strikes have been, like most strikes, demands for better pay and benefits. But they have also, to some large degree, been for what we, might call PMC-specific demands — the right to do one’s job properly, according to its own objective standards. Teachers want to be able to teach.

I think there is the possibility of a broader program here. I think the specific interests and experiences of professional-class workers can be generalized. I think there is a way that they, just like the interests and experiences of industrial workers, can represent society as a whole.

Autonomy in production may be a defining characteristic of the profession class, but it’s something that exists in all kinds of work to different degrees. Anybody who has a job that involves producing some concrete use value can weigh the standards implicit in that use-value, against whatever the bosses say.

The other day, for instance, a  garbageman showed up at our house, a Brooklyn sanitation worker straight out of central casting. We’d put out an old mattress and bedspring. Apparently we had not followed the relevant disposal rules.

“It’s supposed to be completely covered in plastic,” the sanitation guy said.  “The whole thing, like with Saran Wrap.”

Really?” I asked.

“Really,” he said. And then: “Ah, yeah, it’s a stupid rule. I’ll take it the way it is.”

The six year old was delighted to watch as the mattress and bedspring disappeared into the truck’s hydraulic press. Crunch, crrrrrunch, CRUNCH!

Nachruf auf Lothar König: Pfarrer und Punk, 1954–2024.

GIMPS Discovers Largest Known Prime Number: 2^136,279,841-1, first new Mersenne Prime found using GPUs.

Formalising Graphs with Coinduction, by Donnacha Oisín Kidney and Nicolas Wu (POPL 2025). “We argue in the paper that graphs are naturally coinductive, rather than inductive, and that many of the problems with graphs in functional languages go away once you give up on induction and pattern-matching, and embrace the coinductive way of doing things.”

The Design of a Self-Compiling C Transpiler Targeting POSIX Shell, SLE 2024 paper by Laurent Huberdeau, Cassandre Hamel, Stefan Monnier, and Marc Feeley.

getopt() but smaller, neat version by Tony Finch.

“Reflections on Trusting Trust”, but completely by accident this time, by duk.

Neural Networks (MNIST inference) on the “3-cent” Microcontroller, a PMS150C.

A line map for Milano’s trams in TeX, by terru.

One can choose to focus on the car crash, or the lessons learned from the car crash.

Let’s do a little of both.

The proposition of the Living Computer Museum was initially simple, and rather amusing in a Slashdot-baity sort of way: You could apply to get an account on a real, actual ancient Mainframe hooked up to the Internet, which meant you could literally connect into real, actual ancient hardware. I assure you that to a segment of the population, this is an irresistible proposition. It’s also, ultimately, one that even the most ardent fans of “how it was” will leaf away from, because mainframes are their own wacky old world, like using a taffy-pull hook, and appeal on a day-to-day basis to a relative handful of die-hards.

But in 2011, the Living Computer Museum announced itself with the kind of slick webpage that promised computer history buffs a new wonderland.

There’s a lot to take in with this verbiage, but let’s keep going, for now.

In 2012, the Living Computer Museum opened its actual, physical doors to the public in Seattle. A year later, I visited. It was rather nice. I got a grand and lovely tour by some very polite and friendly people, some of whom I’d known for years. The rooms, well appointed, well-lit, and in the case of the machine rooms, done to a sparkling arrangement, clamored for my attention and approval.

There were computer museums for years, decades before Living Computer Museum, scattered among the United States and the world. I’ve been to many of them – sometimes as an honored guest and backstage VIP, and sometimes just because I hastily read my “find me weirdo geek stuff” Google Maps results and negotiated public transit to walk into the door and walk around like a strutting mayor to see what’s what. In that pantheon, I would put LCM in the realm of “very well appointed, especially the mainframes” but not in the realm of “nobody, ever, has ever tried anything like this”. Just off the top of my head, the Computer History Museum in Mountain View has brought not just mainframes and historical computers back from oblivion, but hosted events in which their maintainers and figures have gotten a chance to get their stories on record.

Still, one couldn’t argue with it – the LCM was a solid new outpost in Telling Computer History, and it was on my shortlist of potential homes for materials I might donate in the future. I had a lovely conversation with a member of management about how they maintained functionality and also what their contingencies were for any sort of “endgame” that might befall the endeavor. He assured me they had storage aplenty, in the extremely unlikely event they had to face a closure. Ultimately I didn’t donate my materials there, simply because I found nearer geographic or mission-aligning homes.

My next, and it turns out last, major visit to the Living Computer Museum (outside of a few drive-bys) came in 2018, when I spoke at an event being held there. This gifted me with a chance to see how things had changed in 5 years, and how they had.

Here, then, was something approaching a dream. We had the display cases of technology and the clean carpets and cathedral ceilings of available space, filled not just with computers on desks but entire computer-related environments: classroom, arcade, basement, that showed youngsters what sort of context these computers had lived in. I met Cynde Moya, who had been described to me years earlier as a hard-nosed disciplined caretaker of the materials, and who turned out to be everything they said except hard-nosed – a truly dear overseer of things physical within these walls.

This was April 2018. I lived in the world I live all the time: full of hope, equally full of expectation of doom. But hope was winning.

In October of 2018, Paul Allen was dead.

I suppose everyone has opinions on billionaries, their position in the world, what they represent, and maybe almost everyone has, spoken or unspoken, an awareness that the level of money billions represents disengages you, whether you think so or not, from humanity.

I mean, sure, they cry, laugh, stub their toe, wonder if that’s rain coming, get surprised when the killer’s revealed in a well-done noir thriller. But there’s this cloud of wealth that is ever-present and depending on how it is maintained, or manifested, it slowly bleaches out the edges of living until you realize there is a shocking serenity in their countenance, a noise-cancelling blur that surrounds them, because there are always people whose job is to be aware of them and whose job is to ensure the serenity is maintained.

And there’s the sheer spending power of a billion. I’m sure I could toss out a hundred funny examples of the sheer numerical force of a billion dollars. For example, you could have someone buy a 2025 Subaru Outback, itself the rough amount on the lower end (nationally) of a Domino’s Pizza General Manager’s yearly salary before taxes, and push it into a lake. Taking Christmas and the pusher’s birthday off, to run through a billion dollars of Subaru Outbacks would take you 61 solid years.

But that’s not even very accurate. During that 61 years of pushing Subaru Outbacks into a lake (in Neutral, of course, I’m not a monster), the remaining amount of the money not being spent to purchase lake-bound outbacks would actually be MORE. THAN. A. BILLION. You would actually be richer than you started, if you invested it in even the most brain-dead obvious multi-percent-a-year funds, or left it in any bank. That’s because time and space warp around money.

So do people.

In 2018, after getting cancer in 2009 and 1982, Allen got it again, and this third (public) bout with it ended him, at age 65. His net worth at the time of his demise is the kind that publications who make it their business to will guess at, and the guess sat around $20 billion. Everyone has kind of accepted that, but it’s not really important if it’s $50, $15 or $5 billion. It’s a lot of money.

With his billions, over the decades, Allen owned or partly owned three sports teams, at least 10 companies ranging from airplanes, scientific research, media, and space flight. He owned massive real estate, a movie theater, threw buckets of cash at schools, the arts, and ultimately, a few museums.

To manage it all, he had a company whose purpose was to manage the money, because he wasn’t going to do it himself. The company’s name is Vulcan, Inc. Even after divestment and shutdown after Allen’s death, it has 700 employees. It is one of the largest trusts in the world. It is also now called Vale Group, but I’m going to keep calling it Vulcan.

Vulcan, a company whose job is to manage money and where that goes, is still quite intensely active as an entity, even with Paul Allen’s death. His sister, Jody Allen, chairs the organization she co-founded with her brother in 1986. This company is absolutely gargantuan in its scope and range, and remember, it is a company that just manages a ton of money – that’s the company’s entire purpose.

I wish to take a short moment to not demonize Jody Allen. As the remaining sibling in charge of this company, functionally working through her brother’s holdings, many of which she clearly had no interest or extant position in, and doing so for six years and counting, can’t be anything else than the strangest mixture of pain and endless complication. The $20 billion she’s now in charge of might soften the blow, but likely not by much.

And I apologize for the whiplash here. I’m not overly interested in going down the paths of describing what Jody should be doing, or what her responsibilities are. I am not privy to whatever deep law and lore and functions buried within Vulcan’s iron heart she and her army of people are dealing with, that six years later they are still slowly divesting organizations. I know she has personally called for cullings but I am not informed about the deepest depths of how the aspects of these billions function.

I want to refocus to the fact that Vulcan is divesting itself of the Living Computer Museum.

I want to refocus to the fact that the Living Computer Museum was never a museum.

Look, don’t jump down my throat about this. I’m as shocked as you are.

I didn’t get some inkling from the phone call I had a decade before about potentially donating to Living Computer Museum. I don’t have some spidey-sense about failure or darkness – I just see it everywhere in everything and I treat every day like it’s the one before they find a lump.

If, before it closed, you had made me stand in the center of the LCM and answer whether it was a museum, I’d have happily held up my top hot and shouted “Why yes! One of the finest in all the realms!”

And for all I’d know, it was a museum. There’s no laws on what a thing can call itself regarding being a museum, exhibition, tour, or display. It’s against the law to take money to attend the museum and you get led to an empty lot, sure. But if something has the vestements and affectation of a museum, and you see the big sign saying Museum out front and you go in and there’s displays and staff and events and meetings, you would certainly think it was a museum.

Turns out it wasn’t.

The auction house Christie’s announced it was doing an auction of some of the best portions of Paul Allen’s estate, first with art, and now it would be selling off technology. To fans and studiers of the Living Computer Museum, these items seemed familiar – some of them were on display in the museum.

I found out that some of these to-be-auctioned items would be on display near where I live, so I hopped in my car, parked grandiosely illegally in front of the building, and let myself in.

On display were all sorts of computer history of the “early major days” variety, ranging from a Xerox Star to an Englebart Mouse and even a Pac-Man machine. All had descriptions, all were held out where you could get very near them assuming the guard wasn’t watching too closely, and all of it with a starry-eyed disco-ball aesthetic indicating you were in a space lounge.

So yes, items from the museum were now going to be sold to the highest bidder in a few months. These were some of them, and likely there would be more.

And then, among the relics, I found Stephen Jones.

Stephen Jones and I go a long way back. He was interviewed by me for the BBS Documentary in 2003. So that’s 20 years I’ve known the guy. At our interview session, he showed me his PXL-2000 and his Delorean, both of which I can assure you are high-end geek possessions of the time. And I found his conversation style charming, and his passion for technical subjects perfect for my film.

Stephen’s primary non-profit endeavor, the SDF Public Access UNIX system, was in full swing back then, and is still around now. “Do cool computer things, but don’t try to turn everything into a financial instrument” is not the motto, but it should be. It represents, to me, a lovely ideal of a truly living computer community, incorporating new technology as it expresses itself being something neat, or worth bringing in. Laptop Amish, if you’re stretching around for an anology, which you shouldn’t be. It’s a computer club. And he’s a big part of it.

He was also my initial “in” for the dish on the now-not-so-Living Computer Museum. We’d gone a long way, and here we were, in Christie’s in New York City, standing among offered-for-auction pieces of a gone billionaire’s museum, and maybe we needed to talk in detail, so we did. And I grilled him.

Let’s go back to that initial 2011 announcement page on the Living Computers Museum website, before it was officially an open building.

In the light of its piecemeal disassembly, a lot of this writing hits different. Here’s what I’ve picked up, from Stephen and others, and which I assume for some readers will be new information.

First, this was always intended to be an actual, non-profit, independent/independent-subsidiary museum. It never became that. It is not that. At best, the Living Computer Museum was a billionaire’s (sorry, multi-billionaire’s) collection of computers and technology.

Think of it another way: Jay Leno has a very famous garage that has many wonderful cars and vehicles in it. It is currently, as of this writing, run by a very competent manager named Bernard Juchli, who has been keeping the collection in line for over 20 years. That is, to say, that Leno has a massive collection of cars, competent staff to take care of it, facilities to show them (he has had a number of documentaries and television shows about the garage) and he could probably even open it to the public for an admission price. But you would be hard pressed to call it a “museum” in any sense beyond “it is a display people see videos about”.

As per the request in that 2011 webpage grab, people donated many computers and pieces of history to go into the museum. They expected, no doubt, that they and their children and their childen’s children could stop by displays and point and talk about the family connection to these items.

That was a mirage, a misunderstanding.

This was Paul’s collection of computers, aided by friends and fans. It was always that. The Living Computer Museum, it turns out, cost millions, over ten million a year, to operate as it was set up, and it never came close to making that amount of income from door sales and t-shirts. It never came close to even figuring out how it would.

It’s perhaps not surprising this misunderstanding could happen, and with what felt like all the time in the world, a mere five years and change of being open to the public might have seemed like a mere revving up for the grand plan of what would come next. But nothing came next. It was a collection you could walk through in a really, really nice display case.

And now it’s time to sell the collection.

Before I end up on a huge rant about what it means to be a museum, let me at least give you some relative good news.

The current rough plan is that the big-ticket items, like this Apple I, will be sold at auction. They’ll make beaucoup bucks, go into private collections or maybe even other institutions, and that’s that for the headliners. But that leaves a lot of other stuff.

SDF has a rough plan to raise funds and digitize software, documentation and other related items that are not part of these big-ticket sales (which will be relatively few) so that unique bits of history that are not headline-grabbing trinkets will, potentially, have a chance at being shared with the public.

And I’ve stepped into it. I’ve offered to take software and documentation and store it at the Internet Archive, and of course to work with SDF on ways to digitize/rip/scan materials and, also, host them at the Internet Archive.

It’s all preliminary, but my card is in the hat. Things are not going to be discarded, if I have anything to do with it. A lot can change, and we’ve already seen expectation vs. outcome with this whole experience, but rest assured, I’ve at least tried to provide one last safety net for the kind of history that easily finds itself gone because nobody steps forward. Internet Archive has stepped forward. Stay tuned to see how that works out.

But you know, what did we learn here? What did we actually come away with in the realm of lessons or teachings about what issues this whole mishegaas has churned up?

I have found, perhaps not to my surprise, that a few people in my orbit are displeased I have punched the dead billionaire, ostensibly about all the good the dead billionaire did. But I wish to point out the dead billionaire did give away all that money and yet still had 20 billion dollars at the end, money that went off and did not do as many nice things as the money given away. So much contributed, tax-free, and yet there’s even more. Perhaps this is not the best place to point out that every billionaire is, in many ways, a failure of society, but maybe we can put a pin in it when we have our first trillionaire, assuming that criticizing a trillionaire is not made illegal at that soonish juncture.

I did have some people say, in not so many words, that of course the Living Computer Museum would collapse once its Master of the House was no longer in charge, and I would take this time to point out that Gordon Bell, co-founder of the Computer History Museum, passed on in May of 2024 and yet the Computer History Museum lives and breathes in his absence. It wasn’t a hard problem.

It’s easy to call me ungrateful, but my ungratefulness comes from the narrative people weave near the power of money, like those myths formed around gods and devils, to explain why things are how they are. I appreciate life and its little whorls and eddies that capture us by surprise and we sink or swim. I do not appreciate acting like someone spending a lot of money on something that slightly amuses them earns them the greatest respect. Especially when, as I do now, I walk among a literal army of people working to clean up the mess left behind, after their march to oblivion.

I’ve been writing Go pretty casually for years – the backends for all of my playgrounds (nginx, dns, memory, more DNS) are written in Go, but many of those projects are just a few hundred lines and I don’t come back to those codebases much.

I thought I more or less understood the basics of the language, but this week I’ve been writing a lot more Go than usual while working on some upgrades to Mess with DNS, and ran into a bug that revealed I was missing a very basic concept!

Then I posted about this on Mastodon and someone linked me to this very cool site (and book) called 100 Go Mistakes and How To Avoid Them by Teiva Harsanyi. It just came out in 2022 so it’s relatively new.

I decided to read through the site to see what else I was missing, and found a couple of other misconceptions I had about Go. I’ll talk about some of the mistakes that jumped out to me the most, but really the whole 100 Go Mistakes site is great and I’d recommend reading it.

Here’s the initial mistake that started me on this journey:

mistake 1: not understanding that structs are copied on assignment

Let’s say we have a struct:

type Thing struct {
    Name string
}

and this code:

thing := Thing{"record"}
other_thing := thing
other_thing.Name = "banana"
fmt.Println(thing)

This prints “record” and not “banana” (play.go.dev link), because thing is copied when you assign it to other_thing.

the problem this caused me: ranges

The bug I spent 2 hours of my life debugging last week was effectively this code (play.go.dev link):

type Thing struct {
  Name string
}
func findThing(things []Thing, name string) *Thing {
  for _, thing := range things {
    if thing.Name == name {
      return &thing
    }
  }
  return nil
}

func main() {
  things := []Thing{Thing{"record"}, Thing{"banana"}}
  thing := findThing(things, "record")
  thing.Name = "gramaphone"
  fmt.Println(things)
}

This prints out [{record} {banana}] – because findThing returned a copy, we didn’t change the name in the original array.

This mistake is #30 in 100 Go Mistakes.

I fixed the bug by changing it to something like this (play.go.dev link), which returns a reference to the item in the array we’re looking for instead of a copy.

func findThing(things []Thing, name string) *Thing {
  for i := range things {
    if things[i].Name == name {
      return &things[i]
    }
  }
  return nil
}

why didn’t I realize this?

When I learned that I was mistaken about how assignment worked in Go I was really taken aback, like – it’s such a basic fact about the language works! If I was wrong about that then what ELSE am I wrong about in Go????

My best guess for what happened is:

1. I’ve heard for my whole life that when you define a function, you need to think about whether its arguments are passed by reference or by value
2. So I’d thought about this in Go, and I knew that if you pass a struct as a value to a function, it gets copied – if you want to pass a reference then you have to pass a pointer
3. But somehow it never occurred to me that you need to think about the same thing for assignments, perhaps because in most of the other languages I use (Python, JS, Java) I think everything is a reference anyway. Except for in Rust, where you do have values that you make copies of but I think most of the time I had to run .clone() explicitly. (though apparently structs will be automatically copied on assignment if the struct implements the Copy trait)
4. Also obviously I just don’t write that much Go so I guess it’s never come up.

mistake 2: side effects appending slices (#25)

When you subset a slice with x[2:3], the original slice and the sub-slice share the same backing array, so if you append to the new slice, it can unintentionally change the old slice:

For example, this code prints [1 2 3 555 5] (code on play.go.dev)

x := []int{1, 2, 3, 4, 5}
y := x[2:3]
y = append(y, 555)
fmt.Println(x)

I don’t think this has ever actually happened to me, but it’s alarming and I’m very happy to know about it.

Apparently you can avoid this problem by changing y := x[2:3] to y := x[2:3:3], which restricts the new slice’s capacity so that appending to it will re-allocate a new slice. Here’s some code on play.go.dev that does that.

mistake 3: not understanding the different types of method receivers (#42)

This one isn’t a “mistake” exactly, but it’s been a source of confusion for me and it’s pretty simple so I’m glad to have it cleared up.

In Go you can declare methods in 2 different ways:

1. func (t Thing) Function() (a “value receiver”)
2. func (t *Thing) Function() (a “pointer receiver”)

My understanding now is that basically:

• If you want the method to mutate the struct t, you need a pointer receiver.
• If you want to make sure the method doesn’t mutate the struct t, use a value receiver.

Explanation #42 has a bunch of other interesting details though. There’s definitely still something I’m missing about value vs pointer receivers (I got a compile error related to them a couple of times in the last week that I still don’t understand), but hopefully I’ll run into that error again soon and I can figure it out.

more interesting things I noticed

Some more notes from 100 Go Mistakes:

• apparently you can name the outputs of your function (#43), though that can have issues (#44) and I’m not sure I want to
• apparently you can put tests in a different package (#90) to ensure that you only use the package’s public interfaces, which seems really useful
• there are a lots of notes about how to use contexts, channels, goroutines, mutexes, sync.WaitGroup, etc. I’m sure I have something to learn about all of those but today is not the day I’m going to learn them.

Also there are some things that have tripped me up in the past, like:

• forgetting the return statement after replying to an HTTP request (#80)
• not realizing the httptest package exists (#88)

this “100 common mistakes” format is great

I really appreciated this “100 common mistakes” format – it made it really easy for me to skim through the mistakes and very quickly mentally classify them into:

1. yep, I know that
2. not interested in that one right now
3. WOW WAIT I DID NOT KNOW THAT, THAT IS VERY USEFUL!!!!

It looks like “100 Common Mistakes” is a series of books from Manning and they also have “100 Java Mistakes” and an upcoming “100 SQL Server Mistakes”.

Also I enjoyed what I’ve read of Effective Python by Brett Slatkin, which has a similar “here are a bunch of short Python style tips” structure where you can quickly skim it and take what’s useful to you. There’s also Effective C++, Effective Java, and probably more.

some other Go resources

other resources I’ve appreciated:

• Go by example for basic syntax
• go.dev/play
• obviously https://pkg.go.dev for documentation about literally everything
• staticcheck seems like a useful linter – for example I just started using it to tell me when I’ve forgotten to handle an error
• apparently golangci-lint includes a bunch of different linters

arXiv:2408.03492v1 Announce Type: new Abstract: In this paper we demonstrate how logic programming systems and Automated first-order logic Theorem Provers (ATPs) can improve the accuracy of Large Language Models (LLMs) for logical reasoning tasks where the baseline performance is given by direct LLM solutions. We first evaluate LLM reasoning on steamroller problems using the PRONTOQA benchmark. We show how accuracy can be improved with a neuro-symbolic architecture where the LLM acts solely as a front-end for translating a given problem into a formal logic language and an automated reasoning engine is called for solving it. However, this approach critically hinges on the correctness of the LLM translation. To assess this translation correctness, we secondly define a framework of syntactic and semantic error categories. We implemented the framework and used it to identify errors that LLMs make in the benchmark domain. Based on these findings, we thirdly extended our method with capabilities for automatically correcting syntactic and semantic errors. For semantic error correction we integrate first-order logic ATPs, which is our main and novel contribution. We demonstrate that this approach reduces semantic errors significantly and further increases the accurracy of LLM logical reasoning.