Well, not a lot to report this month; I've been hard at work on this new project which I'll show off at the end of the month (Deconstruct and PoCSci, and probably a youtube video). It's turning out to be harder than I naively thought (isn't it always? and even taking that into account?) but at this point several things are working and I'm satisfied that it is at least possible, so I'm not giving up! Tonight I have to order some more printed circuit boards because the ones I got, although awesome, are very slightly wrong due to my own sloppy measurements. It's an amazing world we live in where you can get high quality custom PCBs from China in a week, but there are not many weeks left for me to wait for boards and realize I screwed up, so it's time to get this one right!
In post 1154 I mentioned that I had finished "most of the reasonable stuff" in Celeste, and now I've actually finished all of the reasonable stuff and some more possibly unreasonable stuff (all regular berries, b-sides, c-sides). That game continued to hold up for me through the c-sides, although it gets pretty grueling. I feel pretty good at it but then I watch the speedrunners and it's obvious I'm still in amateurville. But, I've had enough. Any new games to recommend?
Have kept up the running every day, now at like 8+ straight months. It's much easier now that the weather is getting nice and (especially) my neck is more or less okay. Not expecting to continue the every-day-no-excuses rule after the one year mark, though!
Thursday was SIGBOVIK 2018, held on April -2, the earliest it has ever occurred, due to various holidays and weekends. I don't have any grand project this time, but I did write one paper and coauthor another. The latter was with the SIGBOVIK elder Jim McCann (and was mostly his work); it's The fluint8 Software Integer Library for processors that only have floating point instructions. My other paper was a basically real (but not that rigorous) analysis of a huge database of academic papers to demonstrate that authors with alphabetically earlier names get more citations. That one's called Academic Advancement Advice: Author Articles as A. A.. It also contains some other related study, like the title words most and least likely to get you cited. I wish I had finished some more things for the conference, but on the other hand it is nice that it has so much independent momentum!
My neck has been getting better so I've been doing some running and physical-universe projects, which are not too interesting to describe here. But I have also some fun stuff underway, in particular for two speaking engagements in Seattle in May: PoCSci which is like UW's version of SIGBOVIK, and Deconstruct which is a more serious—if still tolerant/encouraging of weird stuff—event. The project is going well so far but it's a bit stressful, since you never really know if it will even be possible until you're knee deep in it, you know?
Old man complains about discomfort
(28 Feb at 23:37)
Here we are again!
My neck is still giving me and my arm nerves trouble, though it is not as dire as it was in January. I've tried running a few times when it wasn't hurting too bad, including a 5K (Race for the Cure) while on vacation in Florida. I did reasonably okay, finishing 7th overall (eponymous) and 1st in my age group (old mans) with a time of 21m08s. It's a good reminder that the daily elliptical machine or exercise bike does work OK for keeping me in shape, even if I feel pretty rotten. Hopefully next time I'll report that I got better on my own or have some plan to fix it, because this is NOT COOL!
Did I even do anything else this month? I finally built a table saw sled, which is useful for making e.g. other table saw sleds, and finished the basement shelves project I described in the last post. The Learnfun/Playfun work is coming along, though some stuff (from the 3Dification project) that I expected would be easy to apply turns out to be harder than I thought for my test game (Contra) due to the crazy way that game works. So I've done some generalization of that, which is at least a fun puzzle. Unfortunately it's likely that I'll have this ready for SIGBOVIK, so I gotta think of something else (presumably simpler!) for that deadline, which is in two weeks.
I can heartily recommend Celeste if you like a hard platformer game, as I do. It's familiar gameplay wise, but it does almost everything right, has charming art and story, and just loads of challenges and secrets. I've played it a lot this month and finished most of the reasonable stuff (almost all the berries, hearts, b-sides) in 25 hours and 6688 deaths so far, each delicious. 10/10 would buy again.
I've continued up my streak of "running" every day, now passing the six month mark, although "running" now expanded to include the exercise bike. I was already counting treadmill (seems natural) and elliptical (maybe more questionable but it is clearly running-like), but the bike seems to me like a bit of a departure. Unfortunately it's for a pretty annoying reason, which is that an old malady, this herniated disc of mine, has been acting up something fierce this month. I don't know if I ever posted about this (closest I can find is Post #47, Theme from Spinal Fluid, still good), but basically two of my neck vertebrae are squishing the guts of the disc between too hard, and then the disc guts blow a hole in the disc wall and shoot out guts all over my spinal cord, which is where most of the good stuff happens in the body. It looks like this (or did, in 2012):
Theme from Spinal Impingement
The thing at the left is a slice of the cervical vertebrae, the grey tube to its right is the spinal cord, and the big black blob at 0,0 is herniation ground zero. Anyway I had this thing for a while, mostly just a literal pain in the neck, but for the entirety of this month it's been (I presume) impinging on my arm nerve roots and giving me awful phantom arm pains ("radiculopathy"), which are exacerbated by footfalls. This has pretty much precluded real running, and made pretty much everything else pretty painful, so it's been a bummer month. This is a problem that's likely to be solvable eventually (basically worst case is surgery), so that's good, but also will likely involve weeks or months of annoying and disruptive pain. It's good that I have the exercise streak going, because otherwise it would be a very tempting excuse to just laze around.
On the other hand I have gotten some good project work done this month. One is a fairly boring house project, tearing out some old moldy fiberglass insulation in this crawlspace/closet and putting in better moisture-resistant stuff, and then building these needlessly sturdy shelves:
I cantilever not butter
It's a bit hard to tell, but they're cantilevered, which requires more work (good) and provides more space (also good) because there's no support on the front side (so for example I can still put big objects on the floor under them). There are a couple of mistakes but I'm pretty happy with it for a crawlspace kinda install. (Also they're not totally done in this photo; I need to do some finishing still.)
Then, since finishing up my last video I've been back at my old NES "AI" projects again. I've been working on it off-and-on for several years, actually, with lots to say, but this month I finally had some breakthroughs that has it performing pretty impressively my test game (namely, winning it in about 96 hours of CPU). That win isn't totally hint-free like my previous work—you could call it "cheating"—so there's more to do before I'll consider this really to be a proper v2. But here's a image from tonight's experiments:
Basically this graphic shows that the 48 hours of tuning is pointless because the parameter does not affect the outcomes any more than luck (which may actually be good news?). But despite the complexity I'm beginning to put together a strategy for making videos about the new pieces, some of which are interesting, maybe targeting SIGBOVIK? And speaking of SIGBOVIK, the 2018 site and call-for-papers is up, so consider making something!
The movie above is another fun project, I hope entertainingly explained, with some twists. It concerns the dismemberment of letters to generalize anagrams, which I call anagraphs. It was presaged in post 1146 which may give you some idea how long it takes me to finally finish such projects. This one lended itself to some interesting computational problems, which I wisely realized would better exist in a completely separate video; it's this:
This latter movie is far less polished and rated C for College, because it's basically a lecture about decidability, specifically two uses of the technique of reduction to prove an undecidability result, and then a decidability result. If you like that kind of stuff, you will probably enjoy it, but if not, you were warned!
Also crossed off the checklist, I 100%ed Super Mario Odyssey and finished some books I was reading, as well as keeping up my streak of running every single day (today was #159). This makes my slate clean to start on new projects, a state that I just love being in, and will likely postpone through the holiday! Happy new year to you!