Caught up on flashcards, did some grammar, did some listening. Current pain point is discrimination between types of requests — e.g. asking permission, vs. asking how to do something, vs. asking for a favor — and request direction/actors — e.g. asking to borrow something, vs. asking someone else to lend you something. All the forms are familiar. How they're combined, and in which context, is new.
Realized this evening that I was so busy Thursday reading papers on context collapse and messing with Tertulia's multi-user system I forgot to do any flashcards!
Played catch-up today, but it's late and there are still 75 due. I'll handle them tomorrow. I did learn at least one new, fun word: 掲示板 — bulletin board! That one's stuck. I've completed maybe 10-15% of 新完全マスター聴解. Definitely worried during the 問題紹介 that I was doomed. It hasn't been as bad as I thought, given the initial challenge: my troubles are generally elisions that lead to incorrect verb choices, a few missed set patterns I haven't learned yet (e.g. ～させて＋もらえる？ to ask permission), etc. So far, no super unknown or mysterious grammar, which gives me some hope. I haven't spent as much time in my grammar book this week as I've wanted either. Some days learning Japanese feels like continuously pouring water into a sieve. But my metrics bear witness: I am improving. The only question is whether I'll be ready for the December 6th N3 JLPT or not, or if it'd be better to wait until July's exam.
Many things still in flight in my personal life, some things professionally have finally, finally settled down; hopefully I'll find time to study this weekend.
頑張って！おやすみ。
I'm sitting on a full multi-user version of Tertulia. It's off in a branch I've been riffing on with Claude Code.
Every user gets their own journal — a space to think, which is how I use Tertulia myself. Other people can follow you, and you decide how to grant access to the people in your network.
That design comes out of thinking seriously about context collapse after living through it on Facebook. For me, most of that came around the Israel/Palestine conflict: experiences that dragged my entire network into debates over the definition of antisemitism. Conversations a lot of people didn't feel safe having, because they saw only downside — real social and professional risk — and no upside beyond signaling that you're against crimes against humanity.
I've come around to the view that Facebook's privacy controls and friend lists are actually well designed. The problem is what they're asked to do: contain conflicts the engagement systems are built to amplify, on a network you've assembled over years without ever thinking about trust boundaries or algorithmic amplification — and with no good way to move people between trust groups once a conflict has already started. I have 2,400+ "friends" accumulated over fifteen years. How was I ever going to sort them?
Because here's what it looks like when a conflict arrives before you have: every Jewish person in your network — and the whole world — sees your posts criticizing Israel. The people who disagree arrive angry, your friends arrive to defend you, and the feed algorithm decides this is exactly the content it wants more of, because it's "engaging" and "increases time on platform." And then what do you do?
I stopped posting. I decided it wasn't worth feeding the Zuckerbeast if it cost me my sanity, especially when I mostly use social media as a space to get thoughts and ideas out of my head.
I don't know if I'll ever release this thing. But designing my own platform has turned out to be the best way I've found to actually think through any of it.
Thinking a lot about audience management and social media; there's a sort of flattening that happens on FB, where your private posts become the world's fodder; your high school friends bump into your college friends on a post; your new community and its politics ring off in the distance, in a community, hundreds of miles away. I don't want this platform to have addictive features, but I do think it could become a multiuser platform. The hard question is: how do you avoid Facebook's mistakes? What does a social-science informed social media platform look like? Is it worth doing at all if the tradeoff means focusing more on social dynamics than the posting experience, which was the real reason I built this thing at all?
Social media technologies collapse multiple audiences into single contexts, making it difficult for people to use the same techniques online that they do to handle multiplicity in face-to-face conversation. This article investigates how content producers navigate ‘imagined audiences’ on Twitter. We talked with participants who have different types of followings to understand their techniques, including targeting different audiences, concealing subjects, and maintaining authenticity. Some techniques of audience management resemble the practices of ‘micro-celebrity’ and personal branding, both strategic self-commodification. Our model of the networked audience assumes a many-to-many communication through which individuals conceptualize an imagined audience evoked through their tweets.
Americans support Netanyahu arrest by 22-point margin: Survey
https://thehill.com/policy/international/5995104-americans-back-netanyahu-icc-arrest/
I received a really fantastic code review for Apollo, one I've been thinking about all day. It's about `EitherT[IO, Throwable, Option[Entity]]` having two error channels, specifically when using `cats.effect.IO`, and Apollo using them inconsistently. I'll probably post a write-up here soon: resisting the urge to have Claude Code fix it and do it myself.
I was chatting with a friend about his military career. He's late 30s, stellar career. On deck to become a Command Sergeant Major (if he waits long enough). He's asked me point-blank why I haven't become a reservist, or considered a military technology career.
And my honest answer is: they won't take me.
A 22-year-old with a fresh degree in any field is eligible for a commission. I'm not — sixteen years of shipping and securing federal systems, yet the missing credential makes me ineligible at any rank, through any door. The same credential that has never once blocked my career as a technology expert. I served three years at GS-15, senior expert advisor to a Deputy Assistant Secretary at the U.S. Department of Veterans Affairs. Still can't be direct commissioned. Service policy requires a degree to commission at all; and even if a service secretary decided my experience justified an exception, 10 U.S.C. § 12205 states:
No person may be appointed to a grade above the grade of first lieutenant in the Army Reserve, Air Force Reserve, or Marine Corps Reserve or to a grade above the grade of lieutenant (junior grade) in the Navy Reserve, or be federally recognized in a grade above the grade of first lieutenant as a member of the Army National Guard or Air National Guard, unless that person has been awarded a baccalaureate degree by a qualifying educational institution.
So policy closes the front door. And the rank a 16-year professional would actually enter at is precisely the range locked not by policy, but by law. The President of the United States could, at most, make me a second lieutenant, and no power in the Executive Branch could ever promote me past first. And starting at the bottom isn't the insult: plenty of great officers started there. The problem is that it isn't a bottom, but a ceiling. The Army's own direct commissions bring people with my experience in at captain to major, because that's where the work is. Their materials provide guidance:
- five to twenty years of documented technical experience, Python or C
- "emphasis on artificial intelligence and machine learning preferred."
Experience sets the rank — they calculate your constructive service credit and pin captain or major on you based on your number of years of experience. The degree sets nothing, except whether they're allowed to look at you at all. It's a checkbox. Commissioning to lieutenant, only to sit in a billet that's ineligible for promotion, isn't a career path, but little more than a favor to the government. It isn't a fair deal, and it should be fixed.
And there is a fix: amend § 12205.