Tertulia

Social media addiction is very likely real. Not, like, having a website and writing (like I do on this platform), but having a feed algorithm that spins like a slot machine to schedule dopamine hits (rewards) and keep the user on the platform. Or a notification system that lights up constantly, and when you check your notifications, usually there's nothing there, but sometimes there's a notification from someone you actually want to hear from. It's the same neurotransmitter hacking that makes slot machines work, and gambling addicts not simply people who can't control themselves, but victims of the exploitation of their particular neurology. In nature, going back to a place that rewarded you previously was good. Gambling, in the evolutionary sense, paid off in resources: food, water, sex. Social media is simply yet another case where our neurobiology hasn't kept up with our ability to engineer behavior via evolutionary useful, evolved, reward pathways.

There is a very real problem about the actors who are pushing for the current round of legislation to regulate social media companies, ostensibly for "the protection of children." There is merit to the claim that the "addiction" framing gets weaponized to justify surveillance-flavored laws. And there's a separate, sourced case that the TikTok fight in particular is about content: the ban had languished for years on pure China/data-security grounds, then picked up sudden bipartisan urgency in early 2024, right as Mitt Romney was publicly grilling Secretary of State Antony Blinken about why pro-Israel messaging was losing on social media, Marco Rubio was accusing the app of being used to "downplay Hamas terrorism," and Rep. Raja Krishnamoorthi told the Wall Street Journal that Oct. 7 was what moved him to support it. The bill passed the House 352–65 weeks later, folded into a foreign aid package that included Israel funding. Maybe that's coincidence. It doesn't read like one.

This is entirely the reason I've left the mainstream social media platforms myself — in an important way, they're a slot machine. And in the worst way, they're manipulation machines built to amplify whatever's good for a government facing an ICC arrest warrant for crimes against humanity.

先週「新完全マスター聴解 日本語能力試験Ｎ３」を買った。毎日15分ぐらい勉強する予定だ。12月6日の試験に向けて、頑張っている！

www.3anet.co.jp

Shin Kanzen Masuta Chokai Nihongo Noryoku Shiken N3 | スリーエーネットワーク

This book is aimed at those wishing to take the N3 level of the Japanese Language Proficiency Test, and is specifically

make posting on the internet fun again!

Fable 5's UI/UX skills are genuinely impressive.

One of my clients has a few ancient apps they want to modernize. They're data-heavy. The UIs are "a product of their time," so to say.

The trouble is that they're afraid to adopt coding assistants. Lots of proprietary data. Huge amounts of risk in handing that data over to an AI company, or anybody. Risk for developers, too, of using tools that their organizations haven't approved.

But the reality is that there are acceptable, relatively safe ways to get the benefits of coding assistants, without allowing them direct access to your code, or your database.

I've talked with engineers who prompt LLMs around the problem, until they get something that can be mapped onto their real domain. Solutions that work on sample or synthetic data, that can easily be repurposed on real data. Apps that are difficult to modernize often have tangled up UI and data access, and their code is frequently, to use a technical term of art, spaghetti. Brittle, if you prefer. Using the fiction can get you to where you want to go, with most of the benefits of coding assistant workflows, without exposing any of it.

So I guess my message here isn't "trust the AI," but careful boundaries around the LLM and your actual confidential data/system can protect confidential information, you can use better tools, and get their benefits, even in a world where you can't use coding assistants quite yet.

experimenting with diy behavior design. there are a few habits i want to change, like drinking too many cans of soda or going to the café too often, that i want to reduce. nothing i could bring to a doctor and say, "can you help me with this addiction?"

battle plan: i'm pricing specific daily choices. good habits, e.g. not spending $8 on coffee = i reward myself to my travel fund. bad habits, e.g. drinking a can of soda = i withdraw $2 from it.

i've been tweaking the values too. when i considered lowering the reward, my brain yelled at me. good sign i've accurately priced the dopamine.

future me gets my vacation paid for, present me gets better habits and a fun lesson in contingency management, loss aversion, and contemporary cognitive behavioral therapy.

upbeat corporate background music for business and companies (no copyright), some melodies never age

👀 Pro-Israel lobbyists on the negotiating team with Iran? What could go wrong?

(If the choice for how I spend my free time is tinker with an open source library or study Japanese, 日本語を勉強している。Maybe one day I'll get paid to work on a Scala app that requires Apollo. 'til then...)

Claude Code helped me do something that would have taken me days without it, that I'd delayed for six months. 🤷