This Ain't No Content

Expertise and History

2026-01-19T00:00:00Z

It’s odd: I’m not a startup founder or indeed any kind of manager, but I enjoy the Common Cog blog. Recently, they posted an article that should have retained its original title, which I believe was Stop Worrying About Survivorship Bias. (They ruined it by appending “With This One Weird Trick” at the end.) The context is reading histories and biographies about business and famous business operators. I think it’s worth reading in full, even though it’s long—or at least I think so.

Histories and biographies relating to business deal almost exclusively with successes or with unusual failures. But many—maybe most—businesses fail for boring reasons and are never written about. The writing therefore suffers from “survivorship bias.”

The main body of the article discusses expertise and how its study evolved starting in the 1980s. I had been aware of the work up to that point, beginning with de Groot’s study of master chess players in the 1940s. I was aware of Ericsson’s concept of deliberate practice. I was not aware of the idea of ill-structured domains, which this article introduced me to. It claims (plausibly) that business is such a domain, which helps justify reading history and biographies even though they do not reflect the field of business at large. (That, and the fact that as practitioners we don’t want an all-encompassing theory of the domain—only enough to be successful in it.)

Medicine is called out as another ill-structured domain, but I think there are many others, including the one I am most interested in professionally: software development. If true, this explains a lot. For one thing, we’ve had any number of attempts to impose paradigms (to name a few: Structured Programming, Object-Oriented Programming, Functional Programming, Agile Development, DevOps), each initially embraced as the Next Big Thing that would make everything better, only to be discovered later to be, at best, a way of avoiding some of the mistakes made before. In an ill-structured domain, there are no universal truths—or at least few really interesting universal truths—only things that sometimes work, depending on context.

To develop expertise in an ill-structured domain, you need to be able to adapt to novel situations by recognizing patterns you’ve seen before and understanding how to fit them together in new ways. Personally, I think of this as training our own neural network, and similar considerations apply. Fortunately, the human neural network is able to learn from far fewer examples than the artificial sort.

This doesn’t mean I’m going to start reading business biographies. Most are closer to hagiography. Instead, it means I’m going to read widely – about business as well as other subjects. The idea is that valuable expertise in the future will be less about nuts-and-bolts knowledge of particular technologies, techniques, or frameworks (to the extent that it ever was) and more about understanding overall context, and when and how to apply those tools.

One last supposition: most areas where people will still have jobs in the future – other than ones where our bodies are useful in ways that robots generally are not – will be ill-defined domains. It’s been true for a long time, maybe even 60 years, that “routine” jobs have been automated away. That same trend is going to affect jobs requiring specialist knowledge, especially where “specialist” still amounts to something like book learning, even if the books are too advanced for most humans.

Books Read in 2025

2025-12-26T00:00:00Z

This was the year I resumed reading in earnest, mostly ebooks borrowed via the city library. Books that stick with me tend to be psychologically realistic and, if they include fantasy elements, they do not require the reader to disentangle a brand-new society from scratch. Fiction keeps giving me small insights into how other people think and feel.

Here are some of the books I enjoyed.

Long Bright River — Liz Moore

This is one of those where I was shaking my head. It is a police procedural set in a gritty part of Philadelphia, where things are rarely fair. I found the characters appealing (the main character, a beat cop with plenty of experience but low self esteem) or appalling (most others in the book). We get a lot of background on the main character to explain how she became who she is.

The Memory Police — Yoko Ogawa

A whimsical, melancholy book that seems like it should have a political or social message, but the point is unclear. The police of the title are real here, and they do take people away to uncertain fates for thought crimes, but we are never told why. Throughout, things disappear at random and are forgotten. There is a novel inside the novel about a woman who becomes her typing tutor’s lover. The writer of the novel inside the novel seems to be in the same relationship to her character as we are to her: slightly confused and concerned. It is not even clear which of the two novels is an allegory for the other and what in our world might be represented.

Babel — R. F. Kuang

An unfair summary: take Harry Potter, set it in the 19th century, and make the main characters children taken from places that suffered under European colonialism. It is, at heart, a campus novel set in Oxford University, just like the Harry Potter books are boarding school novels. Language and translation have been turned into magical powers activated with silver. The allegory is not subtle, but I was won over by the characters, especially the main character, a Chinese boy forced to masquerade, somewhat, as an Englishman.

The Ministry of Time — Kaliane Bradley

Above average for the genre of time travel novels. In a near-future Britain, a time machine is discovered and used to retrieve some individuals from the past, for purposes that are not fully clear. The main character is a woman with Cambodian parents, assigned to be a minder of sorts of one of the time travellers. Twists and turns, a dash of romance, a bit of sex and violence, and a surprise or two.

Chemistry — Weike Wang

What happens when you fail to complete your PhD and (it seems) lose your boyfriend because he completed his? What if your parents are Chinese and therefore have certain expectations that you hate not living up to? Surprisingly funny, given the sad premise. We also get to see the main character dig her way out of the mess.

Nuclear War: A Scenario — Annie Jacobsen

A book that does not shy away from showing bad things that could happen. Really, the idea that we should have all of these weapons and never use them seems bonkers, but then to use them is even more bonkers, for reasons that are explored in detail.

The Winner — Teddy Wayne

A routine thriller, but it worked for me. It is as much about social class as crime. Our humble-born, social climbing hero is a recent law school graduate hired to a tennis pro in a gated community. His employers are upper class Northeastern WASPs, accustomed to getting their way and not too concerned about what happens to the hired help. He, on the other hand, has to make what he can of the opportunities presented, and not get dragged into anything that could risk his reputation.

Havoc — Christopher Bollen

One of the most unappealing main characters you could imagine, an older woman holed up during the Pandemic in a hotel in Egypt, where Covid restrictions are notably lax. She has a habit of interfering with relationships, “liberating” younger women from their husbands by creating forged evidence of infidelity. This time, circumstances lead her further down the path than she wanted.

Circe — Madeline Miller

A re-imagination of Greek myth through the life of one of the minor goddesses. The deities in general are an arrogant lot, but the male gods are mostly insufferable. Cameos by some of the big names, but mostly we see Circe in her exile. Of course, we are on her side all the way.

The Demon of Unrest — Erik Larson

Another Civil War book, but it is not a genre I have read much of. The Southern rebels are not spared, but the Northerners also make mistakes. There is maybe no way the war could have been avoided, but the North, meaning the federal government, certainly underestimated the problem and failed to prepare.

The Last Samurai — Helen DeWitt

Probably the most challenging read of the year, but also one of the most rewarding. Funny in a biting, sarcastic way. The author is not at all shy about showing off her own learning. That’s risky, but it worked for me. None of it has to do with samurai, or even Japan, except indirectly.

There’s a couple of books I want to mention that I did not enjoy. In fact I didn’t finish either of them.

On the Calculation of Volume (Book I) — Solvej Balle

Think of “Groundhog Day” as a book, but with a female protagonist and a setting in France. In her case, some things are not restored at the start of each day, specifically the things she herself consumes. Otherwise, it is the same, but this character is a lot more earnest and good hearted than the one in “Groundhog”. In my case, I found it too dry, with not enough to sustain my interest over hundreds of pages. (Also, the idea that this was just the first book of a series opened up an awful vista of tedium.)

The Sportswriter — Richard Ford

Very much a book of its time, which was the eighties. The sportswriter of the title is possibly overeducated and underemployed and semi-detached from his girlfriend. Not enough happening for me to keep going.

Experiences With Analyzing My Chess Games

2025-12-19T00:00:00Z

From time to time, I play games on Chessiverse, a site that features bots that play like humans. I like to play without a clock. I also like to analyze the games I play, like this one.

I can load it into scid, a nice tool that works on Linux, and has an option to analyze with Stockfish. This produces a output like the following.

Counting, I see one blunder, two mistakes and two inaccuracies by me and a lot of noise. The lines calculated seem to go nine moves deep, the “+/-” notation means something diferent from “-+”, but I’m not 100% sure. It’s too much.

The analysis tools on the top chess web sites are friendlier, but also include a lot of distractions. For myself, I just want to find the places where making the right move could have changed the result of the game. I’m not a newbie, but also not a strong player, so I usually have at least one bad blunder that I should try to learn from.

The review tool I wrote requires a bit of setup, but once it’s running, I can paste the PGN and get a minimal analysis. In the case of this game, that means I went through the moves (or fast forward using the “Next mistake” button) until I got to this position, where it’s move 16 with White to move.

The tool told me:

16.Rd3 played. Try a better move for white.

It took me a few guesses to find a good move and even longer to figure out why I was going wrong. It turns out to be a frequent blind spot of mine: I will often neglect to think through what my opponent is trying to do. Here, my opponent has just moved their king out of check in response to my move of my knight from c3 to b5. I moved that knight because my opponent chased it away with b4. I should have thought “why is my opponent forcing my knight away?”, the answer to which is obvious in retrospect. That knight was guarding the e2 square and my opponent wants to play Ne2+ next, forking my king and rook. I don’t want to trade away my opponent’s knight for any of my pieces that could take it (knight, bishop or rook), because they’re all valuable. Moving the dark square bishop brings it to a safe square where it guards the critical e2 square. I can afford to drop the f3 pawn. ( 16.Bc4 Nxf3+ 17.Kg2 Nd4 18.c3 bxc3 19.bxc3 Nxb5 20.axb5 would be fine, even winning.)

ETA The main advantage of the tool is that is shows fewer errors so that you can hopefully concentrate on the most significant ones. As this is based on the engine evaluation, it’s not going to find anything subtle. If you get ground down in a positional slugfest, this tool most likely will have little to offer.

I will say that in the ideal case, I should first go through the game myself, without any assistance, highlighting what I consider the highlights or (when I lose) the lowlights. Also, I should try to re-create what I was thinking about, what threats I was worried about, what attack I was hoping to build up to, etc. The idea is definitely, once the top mistakes are surfaced, to understand better how they happened and come up with ideas for preventing the same thing happening again. So I would say it is good to deploy the tool as the second phase of analysis.

By the way, I want to give a shoutout to chess-image-generator, which proved the easiest way to generate the image used above. I needed to install some dependencies before installing it.

sudo apt-get install build-essential libcairo2-dev libpango1.0-dev libjpeg-dev libgif-dev librsvg2-dev

var ChessImageGenerator = require('chess-image-generator');

var imageGenerator = new ChessImageGenerator();
var fen = 'r4b1r/p3nBpp/1k3p2/1Np1p3/Pp1nP3/4BP2/1PP2P1P/2RR2K1 w - - 2 16';
imageGenerator.loadFEN(fen);
imageGenerator.generatePNG('position-at-move-16.png');

Being Human

2025-05-16T00:00:00Z

This is prompted by another blog post, Thoughts on thinking, which surprised me. It’s short, so you should read it rather than trust my summary. But basically, he thinks that LLMs write better than him and, because he uses writing to think, he feels his ability to think rigorously is atrophying. That is a scary thought.

Why am I surprised? Because he says he’s being using LLMs to explore new thoughts and ideas for the last couple of years. It doesn’t match how I’ve beeing using them. To me, the single biggest use of LLMs outside coding (and that is 95% on home projects) is helping out when I can’t quite remember enough to help my son with his high school maths. It’s great for that. Of course, you have to use it judiciously, as if it were a particularly kind and patient tutor, because the risk is that it will just do your homework for you and you learn nothing.

This is similar to the risks the fine blog post talks to, except that he’s talking about his own thinking. I’ll have to be honest and say it never occurred to me to use an LLM that way. But, in any case, I can see the danger by analogy.

There’s also an analogy I remember from my own highschool days. I once did a mock Leaving Cert exam in maths and got an A-. The mock used real exam questions from previous years and I had studied them. In a phenomenon I have since seen described as overfitting, I trained myself too specifically on those questions and not enough on the surrounding context. I got a C in the real exam a few months later.

I want to quote the last paragraph:

And I’m still stuck. But at least I’m here, writing this, and conveying my raw thoughts directly into your brain. And that means something, I think, even though an AI could probably have written this post far more quickly, eloquently, and concisely. It’s horrifying.

Yes, the AI would have written it far more quickly. I really doubt it would have written more eloquently. LLMs are good mimics, but they are not human. I’ll admit, I haven’t tried to get an LLM to write an essay for me, but I haven’t been too impressed with samples that I’ve seen other people publish. It’s much more obvious with AI artwork that some people use to illustrate their articles. It is superficially better quality than what they could have done before, yes, but not as good as a real artist would have produced. (I suspect that a real artist could produce AI work that would be indistinguishable from the non-AI version, but that’s not what we’re talking about.)

In a way, I’m not disagreeing with him at all. I do think that personal voices matter. It’s why I have a blog and I think more people should. (I do see the irony of this every time I look at who reads my work – it is almost all bots, many of them bots that feed LLMs.)

Finishing Off a Game

2025-03-18T00:00:00Z

I like playing through games and I’ve discovered something surprising, at least to me. It often happens that, if I start with end position from a master game, where one player resigned, I cannot win that winning position against an engine. (Maybe it’s not that surprising – Stockfish is famous for being able to wriggle out of tight spots.)

From this observation comes a form of practice that uses an engine to good effect – start at the end of the game and beat the engine. Here’s a way that works for me. It assumes you have LucasChess, an old-fashioned but useful chess learning tool.

Choose a game. I like older games, typically from the 19th century. There are lots and lots of places online that have them, for example chessgames.com.
Skip to the end of the game and copy the FEN.
Start LucasChess
Choose “Play against an engine”
Choose which side to play with.
Choose an engine. Usually you want the strongest external engine you can get.
Change the start position. (This is where you paste in the FEN.)
Click the big “Accept” button to start playing.

During the game if you make a mistake you will be prompted by the tutor and you can either try again or, if you really don’t get it, you can choose the engine move. Usually you want to avoid peaking. Hint: if you find you made a mistake, it may be worth spending some time trying to analyze why you made that mistake.

I have a pet theory that lots of beginning and intermediate players don’t spend a lot of time on this part of the game because they assume the position is winning. Actually proving it, though, is a good test of skill and a reminder that there is still much to learn.

Post-Constitutional

2025-03-04T00:00:00Z

I’ve been thinking on and off about politics, because how can one avoid it? The whole MAGA way of governing looks like a circus. But surely some of the people behind it must have some coherent ideas. I tried to read an article by Russ Vought, Statesmanship in a post-Constitutional moment, for some clues.

I think a critical paragraph is this one:

And in America, the scary part is that this regime is now increasingly arrayed against the American people. It is both woke and weaponized. The national security state, with organs like the FBI, NSA, and CIA, are aligned against the American people, who are outraged by this revolution they never assented to. The FBI is investigating concerned parents attending open school board meetings as domestic terrorists. They are putting political opponents in jail. The NSA is surveilling the conversations of citizens.

In case it’s not clear, this is a reference to the situation a few years ago. It’s the entire justification for overturning the political order and starting afresh. But really? I guess it’s really true that Jan 6 rioters were jailed, but I would like to hear more about what the NSA and the CIA were doing against the American people.

Let’s grant, arguendo, the national security state was weaponized. The Jan 6 participants got their pardons. (Wait, did they all deserve pardons?) Can we expect a pardon for Snowden? Is there anything else that he has in mind to prevent future weaponization of the national security state? We have to assume not every bad actor will be a minion of Joe Biden.

But maybe Mr. Vought’s bigger concern has to do with the powers of the executive branch. If I understand it, his thesis is that the consitutional order of the country has curbed the power of the executive too much for over two hundred years, going all the way back to the Madison v. Marbury case that first established that courts could nullify laws and actions they deem unconstitutional.

There is an aside that is hard to parse but has alarming implications.

The death of Queen Elizabeth II has given us a vivid picture of where the Left is trying to go. An executive in a monarch with massive historic grandeur and symbolism and no remaining authority. A parliament where a Cabinet of MPs head the departments of state, staffed by career civil servants with purported expertise. The Left in the U.S. doesn’t want an energetic president with the power to bend the executive branch to the will of the American people. They don’t want a vibrant Congress where great questions are debated and decided in front of the American people and the tradeoffs made there. They don’t want empowered members. They want discouraged and bored back benchers. They want all-empowered career “experts” like Tony Fauci to wield power behind the curtains.

What to make of this? The British system is just radically different from the American, not by coincidence. It has its weaknesses, but an inability to bend the executive branch to the will of the British people does not seem to be among them. (Does he think that Brexit was an idea cookied up by career civil servants or scheming MPs?) The fact that the leader who matters is the Prime Minister, not the Queen or King, shouldn’t matter unless you’re a royalist nostalgic about the days when the power of a monarch was more absolute.

I honestly don’t get the reference to Fauci and “experts”. He’s not British and what power did he have? He was a spokesperson; he ran the NIAID. He was one of the people behind PEPFAR. (He got a Presidential Medal of Freedom for that.) He edited a medical textbook. Which is the thing is that made him “all-empowered”? Seriously, why is an immunologist with his kind of experience not someone good to have around when you’re faced with a global pandemic?

As far as I can tell, Mr. Vought would like a President to have the powers of a monarch in the 18th century, hardly constrained by the other branches of government, able to rule on a whim. Constitutional restraints that exist today are an encroachment that needs to be turned back. This, he says is how American statesmanship can be defined.

At the end, though, we never do get a clear explanation of how this gets him what he wants. I guess a president who has fewer constraints can write more executive orders to make the government do what he wants. But to what end? In the short term, yes, a lot of government can be destroyed (as can allied countries like Ukraine). In the long term, though, the things that need to be done need to be done, whether you employ people in this agency or that agency to do them. Take for example, the folks at 18F. They did a good job, they got cut, but the government still needs to modernize its technology. That need didn’t go away.

Government Efficency

2025-02-08T00:00:00Z

There is a lot of talk about cutting costs in government and it seems to be focussing on two things:

government employees
foreign aid

I’m a software engineer, so I tend to look at efficiency in terms of performance optimization – if you have a system that doesn’t have the performance (typically latency or throughput) you need, you look for where you optimize it. It’s a big topic for which people write entire books, but the fundamentals are not necessarily difficult to understand.

Typically, you will want to first measure so you understand the problem. Then you will want to look for performance “hotspots”, where a lot of the cost is incurred. These are the places to optimize.

I think the analogy to government spending is fairly obvious: instead of optimizing latency, you are looking at optimizing monetary cost, but the overall principles are similar: you want to look at where the most money is being spent.

Looking at the numbers, it’s important to keep the overall context in mind. The federal budget for fiscal year 2024 was around 9.7 trillion dollars).

Average federal government pay is not bad, a little over $106,000 per year. ¹ According to the same source, there are around 2.4 million such employees, so the total salary cost is a little over 250 billion a year). So salary cost is around 2.6% of government spending.

In FY 2024, the budget for the US Agency for International Development was a little under 24 billion dollars or, alternatively, a little over 0.25% of the total.

If you want to make meaningful saving on government spending, these are not places to invest much effort – you’re only going to make a tiny dent even if you zero them out ². On the other hand, the big ticket items are Medicare and Social Security (each around 16%) and National Defense (around 13% according to the same source).

I understand, these are politically awkward places to cut money. You can go down the list, but then you have to scale back your expected savings.

The numbers here are deliberately rough, but even if they are not quite right, they’re in the right ballpark. The overall point still stands.

(1) These are mostly white collar jobs. Lots of government employees could be making more money in the private sector.

(2) And nobody seems to be seriously arguing that there should be zero foreign aid and zero government employees.

Bookmarking For Myself

2025-01-20T00:00:00Z

Many years ago, I was one of the refugees who left delicious.com (no link, because the site no longer exists) and went to Pinboard. This worked well for many years for my bookmarking needs, which are roughly:

Save my bookmarks
Work from any browser, anywhere
Keep them for the long term

For boring reasons I won’t go into, Pinboard stopped working on that second score. This meant I had a small itch I wanted to scratch, which made it a perfect time to build my own solution. Basically, I wanted the same three requirements, plus a fourth: that it should be self-hosted.

I’m very happy with the result, which leans heavily on a Ruby framework that everyone knows but I don’t want to name because it draws the wrong kind of attention from recruiters. As usual for home projects these days, I also leaned on AI, specifically Claude via Cursor. I think it’s worth writing a few more words on that last subject.

I think I have the same experience as many other people. When it’s good, it can be astonishingly good. I don’t know much about Turbo or Stimulus, but I was able to use them for features because Claude does know them—or at least well enough to do the basics. With the Composer mode, I got close to a one-shot solution for a use case that would have taken me hours to do by hand, especially because I would have to come up to speed on all of the APIs that are new to me.

When it’s bad, it can be pretty bad.

It will get what feels like an idée fixe, where it will insist on re-introducing something I rejected earlier. The advice to stage changes and commit often is important, but you also have to keep a close eye on diffs to make sure there isn’t something unwanted sneaking in.

It can write terribly redundant code, especially when the app gets just a little bigger and it “forgets” what it did earlier. Context in general can be a challenge, especially over many days or weeks.

Tests, and fairly thorough tests, are a must. The “tunnel vision” of the AI means that it is apt to break things unintentionally, and testing by hand takes too much time and, as well as being boring, it is cumbersome.

That said, I’m still happy overall. I have a bookmarking app that suits my needs and that I host myself. It took me a few hours but, most importantly, it was fun, like the hacking I remember doing when I was a kid. The result, which I’m not publicizing widely, is at Bookmarks, a name that is also deliberately chosen to be boring. My motivation here is not to impress people with my tech skills but, as I mentioned before, to scratch my own itch.

Guesing Beyond the End of a Game

2024-12-29T00:00:00Z

I’ve been plugging away at Chess Guesser, based on what I’d want from a tool like this. A lot of the changes are tweaks and bug fixes that I won’t go into.

One of the things I have found is that I can’t always beat an engine starting from the end of one of the games I’m looking at. I think this is an important skill: even an utterly winning position can still be thrown away. Also, at lower levels, players will often make you demonstrate you know how to win, even if they don’t have the patience, never mind the skill, of an engine.

The new functionality changes how the site behaves at the end of a game as recorded.

unlike before, you can keep guessing, right until checkmate
an “engine move” button is enabled, which plays the engine move
you can change your mind about moves, backtrack and try alternatives

I hope others find it as worthwhile as I do. If you try it, feel free to drop me a line – contact info is available on the “about” page of the site.

Making an Epson Scanner Work Under Ubuntu

2024-11-21T00:00:00Z

We had a HP copier/scanner that we stopped buying ink for and used solely as a scanner for many years and this month it finally gave up the ghost. The power button no longer works and I don’t have the patience or the expertise to revive it. So I ordered a plausible looking scanner at an online retailer (you can guess which) that offered to deliver it next day.

The scanner was an Epson Perfection V39 II. (I kid you not, this is the actual model number.) I figured it would be about as hard as the old scanner to use: in that case, it worked out-of-the-box. Not quite true for the Epson. I tried just connecting the USB and running Ubuntu’s default Document Scanner. Nothing, the software just didn’t see there was any scanner present.

There’s no Linux driver available on the site named in the printed instructions (https://epson.com/support), only Windows and macOS. However, if I searched on the site for “Linux”, I could find a page about Linux Support for Epson Products, which links to another page where it’s possible to search for drivers.

I found the best match and downloaded, ending up with a file named epsonscan2-bundle-6.7.70.0.x86_64.deb.tar.gz, which I could unpack easily enough:

tar zxvf epsonscan2-bundle-6.7.70.0.x86_64.deb.tar.gz
# creates a directory named epsonscan2-bundle-6.7.70.0.x86_64.deb
cd epsonscan2-bundle-6.7.70.0.x86_64.deb
# here there is a file called install.sh -- run it
./install.sh
# enter your password for sudo -- installing the driver software needs root privileges

This installs both the proprietary driver and a program called, naturally, epsonscan2. It did not install all the dependencies I needed. In particular, it didn’t install some QT libraries, causing this error message:

epsonscan2: error while loading shared libraries: libQt5Widgets.so.5: cannot open shared object file: No such file or directory

I installed it myself:

# first searching for what deb has the library
apt-file search libQt5Widgets.so.5
libqt5widgets5t64: /usr/lib/x86_64-linux-gnu/libQt5Widgets.so.5
libqt5widgets5t64: /usr/lib/x86_64-linux-gnu/libQt5Widgets.so.5.15
libqt5widgets5t64: /usr/lib/x86_64-linux-gnu/libQt5Widgets.so.5.15.13

# then installing it
sudo apt-get install libqt5widgets5t64

Having done that, was able to run epsonscan2. I ran it from the command line, in order to see any error messages more easily.

I ran into some more error messages trying to run it:

QSocketNotifier: Can only be used with threads started with QThread
libpng warning: iCCP: known incorrect sRGB profile
libpng warning: iCCP: known incorrect sRGB profile
qt.qpa.wayland: Wayland does not support QWindow::requestActivate()

As it turned out, these didn’t prevent it from starting. After some playing around, I connected the scanner while Epson’s program was running, it detected the scanner and initialized the driver correctly. I could hear the scanner make some mechanical noises and see the blue LED light up on the front of it.

I didn’t use epsonscan2 for anything more than the first setup. After that, the regular Document Scanner now works, and it’s far more polished and easy to use. Yes, I can confirm, it works with Ubuntu 24.04.1 LTS (Noble Numbat).

There is actually online documentation for epsonscan2 that you can consult if you run into issues. I found it too late to use it myself, but it might be helpful to someone else.

Introducing Chess Guesser

2024-11-09T00:00:00Z

Chess Guesser is a pretty simple application that I wanted for a while and finally got around to implementing. The aim is to allow you to play through an arbitrary game, guessing the moves that you want to guess, and going forward when you don’t want to guess. There are other sites out there that allow you to do something similar, notably chessgames.com’s “Guess the Move” feature, but I like mine better.

Notable features:

if you guess a move better than the move in the game, you get credit
it doesn’t keep score of which guesses you got right
it allows you to skip to two possibly interesting parts of the game, taking the role of the winning player
it allows you to upload a PGN of your choice

A few words about those “interesting parts”. They’re labelled “First Mistake” and “Critical Moment”. In both cases, they are after the winning player has got a significant advantage, as judged by an engine offline. The “first mistake” is short for “first serious mistake” – it is when the losing player makes a mistake that they never recover from. The “critical moment” is short for “last critical moment” – it’s when the winning player makes a move which makes the biggest evaluation jump.

The source for the application is on Github. Like all my recent personal projects, it benefits from AI assistance, mainly Claude as mediated through Cursor. I have lots of opinions about the AI, but it does get me over the hump of reluctance to take on things that aren’t as much fun. Fortunately or unfortunately, I am opinionated, so I revert if things aren’t going in a good direction. Still, you can find plenty of false starts and mistakes if you look in the published git history. What is there is definitely not perfect. I think the biggest misfeature right now is how well it works on mobile devices, particularly Android. However, it’s beyond the “brown paper bag” phase, so I’m happy for anyone to take a look at it if they’re interested.

First Post

2024-10-20T00:00:00Z

[ tap tap is this thing on? ]

…okay, not first post, but first in a long time.

Really, doing this stuff has become so easy now that I almost feel it is my duty to resurrect the blog. Back in the early aughts, Movable Type was the way to put up a personal blog. In some ways, it held up well. If you look back at the really old articles in the blog, I kept most of the HTML and the styling unaltered. That was before xyz-as-a-service was a big thing, so everyone who did it had to either figure out how to get it working or they paid someone to do it.

What’s better now? Everything. Hosting has become stupid cheap, even a dream (for me) setup of a Linux box with full root access by ssh. TLS support is both free and almost idiot proof, thanks to Let’s Encrypt. There are any number of static site generators. Even the tedious stuff of editing dozens of old html files (the old posts) is made much less painful because of LLMs. (It might be tough to get help in a tool Cursor with challenging programming, but it does boring tasks like scripts to rewrite links without complaint.)

The aim is always to reduce the friction of publishing personal material (not “content”, mind you, the title of the blog is old but the feelings remain) to the point where I could do it every day if I wanted to.

Okay, let’s see if we can publish.