The patron saint

The reception traditions at the alma mater have most likely changed considerably since my undergrad days, but at the time there were certain rites and ceremonies that had to be performed for the recently admitted students to be accepted as full members of the school, the final being a party arranged by the new students for the rest of the section. Of course I ended up on the arranging committee. As the party would coincide with All Saints’ Day, we thought we should do something saint-themed, such as introducing the patron saint of programmers, whoever that might be. So I spent some time trying to get hold of a Catholic priest I had located in the phonebook (yes, it was all analog in those days). When I finally reached the priest, it turned out he belonged to the Liberal Catholic Church which had no truck with saints. Well, I had no idea! By then it was too late to start over with finding a Roman Catholic priest, so we had to drop the saints in favour of other ideas which I hope are safely forgotten by now.

Still, with the advent of the World Wide Web, it is now possible to search for all desired information, and one then finds that the inofficial patron saint of programmers is Isidore of Seville on account of him having compiled the first encyclopaedia, so here he is.

Finished model 2022–III

Another Wood Elf Wayfarer. Painted with Humbrol Enamels—they seem to be the only thing that sticks to the metal surface. Intended for the “Paint it black” theme at C4-Open, but was not considered worthy of judgement.

Eliza’s children

These last few years there’s been all this excitement about artificial intelligence, machine learning, and “algorithms”, but at the end of the day I am rather disappointed. It seems the current approach to AI is all about frequency analysis, feeding in billions and billions of pieces of data and working out the most common correct response to each possible input. This basically means the method will be simply be reproducing what is fed into it.

Back in the day, the idea was to develop learning systems with a world model that they could reason about, but that is hard. The Cyc project has been the most high-profile one and still does not seem to have much overt use. (I find it somehow ironic that a site dedicated to the best in AI uses CAPTCHA to avoid spam in their contact forms, but maybe it’s sufficient to weed out the most stupid scripts, and the smarter ones might be worth conversing with.)

The original conversing agent is ELIZA, still available at a text editor near you. As was observed at the time, people who should have known better were taken in by the seemingly intelligent responses, and the same seems to have happened to Blake Lemoine at Google who got tender emotions towards a piece of software. Now, I have not specifically tried the LaMDA system, but others that are built on similar principles and what has struck me with them is that they do not seem to have a very deep knowledge representation of the conversation itself, but exist in the here and now of the current utterance, much like a senile person might do, reproducing patterns, but not connecting the pieces together.

Bits from a recent conversation with a chat bot that pretends to be a person:
I have a dog and a cat
Simulated ones?
No, they live with me
I work from home, so I have plenty of time for them
[ … ]
Do you have any pets?
Yes, I have two cats and a dog
They keep me very busy. How about you?
Ah, are you making this up as you go? Earlier you said you have one cat and one dog.
I am not making it up
I do have a cat, but she lives at home with my parents
So how many cats in total?
There are four of them
They are all rescues
Exponentially increasing.

It’s an interesting bug that the last sentence in each utterance misses the full stop, but not other punctuation.

Finished model 2022-II

Another horrid paper model: The Royal ship Vasa. Box scale, or rather, post card scale. The double-curved surfaces were impossible for me to bend correctly, I wonder if there is some magic trick to make the card stock more pliable. By the stern there were several very thin wedges the were to be cut and glued to make the curved shape. Only afterwards did I realise I should have just cut away the tiny triangles intended as glueing surfaces and instead glued a wider piece of card onto the back. One may note that the bowsprit, which carried a quite sizeable sail, is not represented in the kit. I could perhaps have used a cocktail stick to add it, but couldn’t be bothered.

Finished model 2022-I

As mentioned long ago Pilot 13/1977 contains 3-plan drawings of the Cavalier Turbo Mustang III, and suggestions for how to build one, based on a P-51D Mustang kit. I thought this was a cool-looking aircraft that should be built at some point. Later, I decided I would build it in 1:160 scale for a train modeller friend of mine. When I found a 10 mm scale Minifigs white metal model of a P-51C Mustang, I thought maybe I could use it as the base for a conversion and started collecting references. A fellow modeller, whose name has been unfairly and unfortunately lost in time, mailed me his entire collection of magazine clippings on the Turbo Mustang III. I designed decals in 1:160 scale and had them printed by Al Superczynski. Then I spent several years angsting over how to do surgery on the metal model. I had studied the Heritage Aviation Models conversion kit and had some ideas, but it seemed a daunting prospect anyway, so in the end nothing happened, until I recently noticed that the Shapeways Marketplace contains a number of aircraft models of varying quality and wwitalik has a Turbo Mustang III model which is printable in 1:160. (The model claims to work as a Piper PA-48 Enforcer as well, which isn’t true—there are numerous differences between the two types, not the least being that the Enforcer’s engine exhaust is on the left side of the fuselage.)

I ordered the model printed in “White Natural Versatile Plastic” and it soon arrived. In order to be printable, several features of the model was grossly over-scale and I did my best to adjust these. The propeller blades and tailwheel were cut off, as they were mostly just plastic lumps. The main landing gear is also overly thick, but I felt uncertain about replacing that with anything sensible, but maybe I should have tried. The trailing edges of wings, stabilizer and fin had to be sanded down, which turned out to be difficult—the plastic is rather elastic and tended to just follow the sanding stick and then left grainy fibres along the edge. In the end I had to call it a day and consider the edges as thinned as I was likely to succeed with. In contrast, drilling out the exhaust was actually quite easy. Another things I left off was milling out the landing gear wells, as I wasn’t convinced I would be able to do it well—I would need some kind of way of carefully steering the cutter. Likewise, I briefly considered heatsmashing a transparent cockpit canopy, but apart from the technical difficulties of that, that would also have required milling out the cockpit without accidentally going through the sidewalls. There is a not-particularly expensive milling attachment for my Dremel, maybe I should invest in that.

I fashioned thin propeller blades out of Contrail profiles, cutting and carefully twisting them in warm (not too hot!) water to get blade shapes, landing gear covers were fashioned out of Plastruct sheet, tail wheel covers out of a slice of some aluminium can I had lying around, and a new tail wheel was made from a bit of Plastruct rod and piano wire. I primed the kit with Mr Surfacer 1200 and eventually started painting it. The Turbo Mustang III was intended as a COIN aircraft and at the time it would have meant deployment in Vietnam, so was painted in the then-current South-East Asia camouflage. This meant Humbrol 28, 116, 117, 118. I tried Maskol, again, to mask the colour fields and it seemed to work better now, but the Tamiya masking tape I used to mask off the bottom colour managed to lift a bit of the paint, and even some primer. With the uneven surface that the Mr Surfacer hadn’t managed to improve, airbrushing didn’t make much of a difference, and since I had to touch up scuffed paint at several points, I finally gave up and just brush painted the lot. The control surface hinge lines were accentuated with a pencil. (I need to get me a 0.3 mm pencil again, it has its uses.)

The result will not win any prices (I tried), but it felt good to finally get the old project out of my system, even if in a somewhat different form than originally envisioned.

Trust issues

Reading student reports I often get side-tracked into closer looks at their sources. Not conducive for quick grading, but one can find lots of interesting things. In this case a student pair had looked at drug abuse statistics at the National Center for Drug Abuse Statistics and, being source-critical, noted that the claim “The front page of DrugAbuseStatistics.org features the most noteworthy data regarding drug abuse” didn’t have any supporting source itself. I had to agree that this was the case. Furthermore, once I started browsing the website I realised there was absolutely no information about who NCDAS are. No address, no “About Us” box, no nothing. It seems unlikely that this would be some government organ.

Next step was to google their name to see if anyone else commented on their affiliation, but here, too, absolutely nothing. However, lots of hits on other sites that refer to the statistics collected by them. And I wondered: Is the official-sounding name enough for everybody and their grandmother to assume that this is the best data available, citing them, thus making sure they get to the top of the google searches for drug statistics? They do give sources for their statistics, but even I haven’t made the effort to double-check that the statistics are correctly quoted from the given sources, whose quality in turn is unclear.

To be sure, they might be perfectly legit, but how do we know that?

I write like lots of people

I Write Like uses statistical analysis to compare your writing style with that of well-known English-language authors. I tried different entries from this blog and each of them was apparently reminiscent of a completely different author. Maybe a longer text would give more dependable results? I tried feeding in my PhD thesis. It was most similar to Agatha Christie. I hadn’t earlier noticed the extensive note apparatus of Murder on the Links. Actually, I didn’t notice it now either. For comparison, I also fed in my licentiate thesis, by an even younger me, presumably less mature. That was most similar to the writings of Gertrude Stein. A matrix row is a row is a row is a row?

But, Siggy explains why the algorithm is so often wrong.