You should have a blog

September 21, 2026 8:30 pm

A blog? Really? What year is it? 2008?

Yes, a blog.

Back at the end of the first decade of the 21st century, when you had that blog you’ve long since forgotten about, you wrote some thoughts and off into the ether they went. Maybe some friends and family read them, maybe not. That part would still be true today, but there’s also something different this time: LLMs and the insatiable demand for data to ingest.

Humans reading your (and these) words is cool. Maybe it sparks new ideas in their mind. Maybe it just keeps them loosely acquainted with your life. They exist and they matter.

And in the before times that’s likely as far as any of your writing would go.

Today, every scrap of data is being grabbed and fed into the LLM training systems. LLMs are defined, in large part, by their “weights.” You can think of these weights, loosely, as how strongly connected certain concepts are within the system. With your words being fed in, your words (and these words) will tweak their weights ever so slightly. And once released, someone using that new LLM will see output affected, if ever so slightly, by your words.

That infinitesimal amount is, for most people, more impact than ever before.

So go on, dust off that blog and share something human, real, and good. They’re certainly getting enough training data of the worst humanity has to offer.

3D Prints August 2026

August 31, 2026 8:15 pm

We have some rollarms for our Lovesac sactionals which have the odd annoyance of being round–so you can’t set a drink on them without risking it toppling over. Lovesac does sell some “coasters” for them for $50 (when on sale), but that’s a bit much for a mild convenience factor. I thought this was a great opportunity for some 3D-printing. I checked for existing models and found many for the regular flat sides, which don’t have nearly as much need for a coaster, but for which designing one is significantly simpler. So it was up to me to provide for the world.

A few rounds of doing test prints (just a sliver of the design, not the whole thing) and tweaking and I’m quite happy with the result. I designed it once in SolveSpace (the design application with which I’m most comfortable), but I really wanted all the edges to be rounded (no sharp corners!). SolveSpace doesn’t handle that very conveniently, but Dune 3D does and I’ve been meaning to give it a try. So, knowing what my end goal looked like I designed it again in Dune 3D and added all the edge rounding.

I’m quite happy with the result. It’s snug and feels secure. I printed it using 3% gyroid infill, but with 6% infill throughout the thin section at the top to provide additional strength since that area is going to receive the most force and provide the snugness by resisting flexing. That may not have been necessary, I don’t know, but it doesn’t add much extra filament. With the rounded corners you’ll want to use adaptive layer heights to help them come out cleanly. All these settings are included in the Bambu Studio 3mf file.

Total cost for the final print was about $4.50. The 3 test prints were about $0.30 a piece.

Lovesac rollarm coasterThe 3mf file (14MB)
Lovesac rollarm coasterThe Exported STEP file (406KB)
Lovesac rollarm coasterThe original Dune 3d File (80KB)

Note: the Dune 3D does not include the reference image I had loaded, so it will warn that the image file can’t be found. Shouldn’t affect the model itself.

Games August 2026

7:46 pm

Played a couple games of Mystic Vale at the board game meetup. Lost both, but it was still fun.

My campaign-games group finished our campaign of Kinfire Chronicles: Night’s Fall with a pair of victories on the final two quests (14 and 15). We defeated the guardian of the holy vault, recovered the tome containing the secrets of kinfire, then climbed the enemy colossus and struck it down before it could destroy the Great Lighthouse of Din’Lux. It was fun. The game is definitely easier with four players than with two and we had some really good complementary decks giving us some nice, high-damage combos.

We’ll be playing Earthborne Rangers next.

Books August 2026

7:29 pm

Driving Honda by Jeffrey Rothfeder

An interesting look inside the history and guiding philosophies of the Honda conglomerate.

For some reason, the author spends the final chapter on an only-tangentially-related diatribe on national manufacturing policy.

Blackett’s War by Stephen Budiansky

A look at the inception of what came to be known as “operations research.” Effectively, the intentional application of scientific principles to processes and decision making–specifically in the context of World War II.

When your merchant ships are being sunk by U-boats, should you group them into convoys or send them alone? Should you use your airplanes to bomb factories or submarines? To get your troops fed faster, should you build another mess hall or reorganize the one you have?

The Admirals by Walter R. Borneman

A group biography of the only five-star Fleet Admirals in U.S. history (thus far). After their backgrounds and early careers we then get a high-level history of WWII in the Pacific theater.

Implicitly covers the transition in naval theory from great fleet-on-fleet engagements won by battleships to standoff battles via carrier groups and projected air power.

Games July 2026

July 31, 2026 8:50 pm

The group completed quests 12 and 13 of our Kinfire Chronicles: Night’s Fall story. Both completed victoriously. Quest 13 was clearly set up as a major battle, but with fortune on our side and complementary decks we were able to deal out 80 damage across the enemy team with just 3 attacks and then mop up the remnants from there.

At a board game meetup I jumped into the tail end of the silly game Hot Streak. The players are compulsive gamblers betting on the outcome of a race of sports team mascots as well as various propositional bets along the way. I lost.

After Hot Streak, we played a game of 7 Wonders. I lost that too.

And closing out that night was a game of the rather odd Frenemy Pastry Party. In this game the only way to earn points is to either bake a cake or help someone else bake a cake. Baking a cake alone is suboptimal as it uses up your own ingredients. Contribute ingredients to someone else’s cake earns you points, but they will earn more points for the cake than you will for the ingredients. So the goal is to try to get all the other players to evenly contribute to your own cakes so the points for ingredients are diffused across the other players and cake points are concentrated on yourself. But once every figures this out cooperation is likely to deteriorate. I lost by a point once another player realized what I was doing and rallied the rest of the table to prevent me from winning.