Skyfi.com is a startup that lets you order a new satellite picture taken like you are a movie government agency, but on the web for $175. I got one for fun, and a day later I had a 7k x 7k image sent to me. It looked good on opening, but when I zoomed in close,\ the pixels looked terrible, with saturation, gamma, blocking and stretching issues. I was pretty sure I knew what most of the problems were, so I got in touch with the developers, and everything got sorted out in a few days (they were already aware of some issues).

This is an example of an “object culling” bug in VR Shell, but it is an error that you can see in a lot of 3D software outside the battle tested engines. You want to avoid wasting time drawing things that won’t be seen, so you typically first do some calculations to see if a \ rough approximation of the object, usually a bounding box, would be off screen. If so, you don’t draw the object. That is straightforward for static objects, but more complicated for animated ones, where just calculating the bounding box can be a lot of work. In this case, \

When I first started getting successful in my 20s, I went through a period where I loaned money to anyone that asked me. Well into six figures went out — “emergency” expenses, start a business, make a down payment, etc. Most of the people just never spoke of it again, and \ I didn’t press them on it. It was a learning experience for me, calibrating some general people behavior. I don’t think anyone intended to not pay me back, but things often just don’t work out the way you plan. To my surprise, nearly 25 years later, one of the larger debts has \

I resigned from Meta, and my internal post got leaked to the press, resulting in some fragmented quotes. Here is the full thing: facebook.com/permalink.php?… As anyone who listens to my unscripted Connect talks knows, I have always been pretty frustrated with how things get done at FB/Meta. Everything necessary for spectacular success is right there, but it doesn't get put together effectively.

I learned from @fabynou ‘s book amazon.com/Book-CP-System… that the arcade system had rarely used, dedicated hardware to render star fields. I have considered that in VR — there is no way to render “perfect” star fields today, because the compositor will distort anything you render\ In such a way that the point light stars will not be the perfect 4 pixel linear distribution regardless of what you do. “I could fix that” lurks in the back of my mind when I see an occasional space game, but I know it would be ridiculous to define and support an interface.

A few years ago, in wildly out of character behavior, I attended three rather exclusive private tech gatherings. I was kind of surprised to find them worthwhile. I met some really admirable people at them, and the interactions were qualitatively different than e-mail/twitter/etc\ \ I feel I benefitted from them, which brings up the question: should I go to more of them? I have a deep mistrust of anything resembling going on a “convention circuit” instead of actually working, and I mourn any work week with a hole shot in it, but, again, I got value there.\

I do not have “magical” discipline and focusing ability — I struggle with distractions like everyone else. Often it is better to just remove the option to distract yourself instead of fighting he urges. When I decided to get serious about my AI work, I made a few changes: \ I have been officially “one day a week” at Meta for a couple years now, but I still wound up checking in on groups and email every day, which was often a distraction when I should be concentrating on AI. Now I put my Meta laptop in the garage at the end of my VR day. \

I went through four @weareFitXR workouts yesterday. It was a good experience, and I am pleasantly sore this morning. A few app comments:

The multiplayer was a big draw, but we wanted to be able to go through multiple workouts as a group without creating new rooms and rejoining.\ With multiple people, it is usually good “gamification” to break the scoring up as much as possible. You can have a grand total, but also displaying the results of each “stage” gives more opportunities for each person to win something, even if not the overall. \

I should write a longer retrospective-with-distance on the alt-space times, but a few thoughts:

https://twitter.com/wikkit/status/1553042158801780736

Many companies were aiming at the suborbital space tourism market. It is interesting to see Blue Origin and Virgin Galactic both spend more money to build that than SpaceX and now others took to make their first expendable orbital vehicle. BO and VG are never going to turn \

I watched Tom Swift. I have enjoyed some CW shows, and I went in with managed expectations, but I didn't find it good enough to watch a second episode.

I have fond memories of reading all the original Tom Swift Jr books when I was a kid. I saw some nods to the source material, \ but it was really more of a "young Tony Stark" play than a Tom Swift play.

I wound up spending more time thinking about a meta point around the show's existence than the show itself. I was only even aware of the show because I saw some twitter outrage, which it seems \

When I let myself spend a day doing CUDA work, I'm usually quite happy afterwards. Optimization is a grand puzzle game with an objective score function! I have to make myself stop at a wise point, despite all the intriguing rabbit holes left to pursue. \ CUDA is not my go-to go-fast tool -- with a modern, big CPU, it is pretty great how much performance you can trivially get with #pragma omp parallel for. The last couple times I have done it, my first cut at a CUDA kernel was slower than the C++ running on 128 threads. \

Id Software very nearly got acquired by Activision, before winding up with Zenimax. Interesting that both roads would have led to Microsoft in the end. @BobbyKotick was always supportive, and it is quite an achievement to lead a company through that much growth over that long.

https://twitter.com/xbox/status/1483431175041347584

Ok, I had a sense that Bobby was not well liked, but I didn’t know he was at full un-person status. I last had dinner with him ten years ago, and the sum total of our interactions aren’t that large, but my first hand personal experience was of a shrewd businessman that led one \

The Internet Archive struggles mightily to save content when services shut down, but it is often a panicky reactive process. I wonder if there could be a world where the IA acts as a default host-of-record for startups, with a super-easy CDN relationship such that the content \ is archived-by-default in a heavily bandwidth restricted way, but your customers are served through conventional commercial means. I could imagine a CDN giving a nice rate for the "public good" of a continuously archived and mirrored service. I have a vague feeling that \

Had a long talk with @ESYudkowsky about AI safety. The pessimists fear that we only have one shot at getting alignment right, and will probably blow it, but I think we will have ample experimental opportunities. \ Our core argument is the power provided by superhuman intelligence.

Being Really Smart isn't going to let you deploy nanotechnology weapons overnight, or immediately hack human psychology to the level of mind control. That may be possible over years, but not days or weeks. \

People like the idea of hard and fast rules, but putting +/-infinity as a factor in a policy decision is almost never the best plan. Hiring is an obvious example, with "requirements" that filter out large chunks of applicants that don't have some kind of back channel \ influence. Disqualifications are rarely explicitly stated, but they exist. If a company has a reasonable flow of applications, just tossing out all the ex-cons is an easy call. On the other hand, applicants often don't appreciate that interviewing prospects is a very non-trivial\

A lot of indie game devs want to do everything themselves, either by leaning on the asset store, or by becoming a polymath coder/artist/modeler/sound designer. It isn't impossible, and everyone has their favorite example, but it definitely isn't the high-probability path to \ actually produce something successful. If it is a personal growth hobby, then fine, but if you want to compete in a very crowded market, expanding the team with complementary skill sets is usually critical. I think about this a lot as I sit here working on AI by myself.

Before the iPhone existed, I worked on a few games for what were called "feature phones": Doom RPG 1&2, Orcs&Elves 1&2, and Wolfenstein RPG.
Qualcomm's native-code BREW platform had better versions, but I haven't seen any emulators and archives for it, so they may be lost at \ this point. The J2ME (java mobile) versions are still floating around, and can be emulated.
My son wanted to get O&E2 running, so we set out on a little adventure.
Kemulator ran the game, but audio was glitchy and it hung after you died in game. Well, we are programmers, we \

I get asked for career advice a lot, and while my "learn deeply" pitch may be good long term advice, it doesn't help breaking in -- being able to write your own tool chain with a hex editor is great and all, but it doesn't add value at most companies. \ I suspect there is a useful path that I think of as the "tool master". Modern art and programming tools are enormously complex systems, and the typical user only touches a tiny fraction of their features. Picking a path of study that revolves around deeply learning a tool rather\

After complaining that numpy took many hours to solve a 64k x 64k matrix, I broke out cuSolver, Nvidia's GPU linear algebra library. A 32k matrix gets solved (LU decomp) over 1000x faster than base numpy (with MKL not loving my AMD CPU), but a 64k matrix of floats is too big \ to solve directly on my 24 GB Titan RTX card. The nice thing about working with a low level library is that you have to explicitly allocate the temporary working buffers, so when it doesn't fit on the device, I can put it in pinned host memory or on my other card connected \

The Imperial College epidemic simulation code that I helped a little on is now public: github.com/mrc-ide/covid-… I am a strong proponent of public code for models that may influence policy, and while this is a "release" rather than a "live" depot, it is a Good Thing. Before the GitHub team started working on the code it was a single 15k line C file that had been worked on for a decade, and some of the functions looked like they were machine translated from Fortran. There are some tropes about academic code that have grains of truth, but \

AMD 3990 CPU scaling tests: Because of the Windows group limit of 64 CPUs, just firing up a lot of C++ std::threads didn't work great:

128 t = 67 s
64 t = 63 s
32 t = 84 s
16 t = 160 s
8 t = 312 s

32 to 64 threads wasn't a big boost, and 64 to 128 was slower. However! \ Setting the group explicitly let it scale all the way up:

128 t = 38 s
64 t = 48 s
32 t = 84 s
16 t = 160 s
8 t = 312 s

Notably, because each group gets 32 hyperthreaded cores, 64 threads across 2 groups on an unloaded system is much faster because they are all alone on a core\