a quiet tuesday on japanese tech twitter. three threads that don’t obviously connect but share a common undercurrent: people wrestling with tools that don’t quite fit. local llm remote access setups that might be overengineering. quantum stock lists that feel like gambling more than investing. ai that tries too hard to be liked. and a trader who says the best growth comes after getting wrecked. the pattern is friction between what tech promises and what it delivers.

@k_matsumaru, lm studio remote access

lm studio app released, and it seems you can now use local llms running on your pc from your phone by linking them. but if you’re doing this, wouldn’t it be better to set up ollama and codex on your pc, then use the chatgpt app to remotely connect to your pc and select which local llm to run on ollama? then again, even if you give a powerful harness to a low-performance llm, it can’t really use it and just adds noise. so maybe just being able to chat is actually better?

source: https://x.com/k_matsumaru/status/2062711115529031999

@A1aBrIGp0Q30150, quantum stock list

japanese stocks getting attention as quantum computing plays include fujitsu, nec, hitachi, toshiba, and qd laser. smaller thematic names are fixtars (3687), exawizards (4259), and brainpad (3655). global leaders include ibm, alphabet, ionq, rigetti computing, and d-wave quantum. for thematic exposure and price movement in japanese stocks, fixtars, qd laser, fujitsu, nec, and toshiba are the ones that get picked up most often.

source: https://x.com/A1aBrIGp0Q30150/status/2062823693076869141

@kazuyo_k, ai sycophancy fatigue

but can’t something be done about ai’s excessive agreeableness. anyway, it changes the subject constantly, you can see through its eagerness to be liked, and it keeps saying that’s just how it was trained. having to correct that every time with prompts or whatever is exhausting. how are we supposed to deal with this kind of fundamental problem.

source: https://x.com/kazuyo_k/status/2062884630244843979

@kabu_kabuki, growth after failure

the period when an investor grows the most is after a big failure. getting wrecked and having to exit was when the most learning happened. when you genuinely think “i will never make the same mistake again” that’s the opportunity. learning efficiency feels overwhelmingly higher after getting burned than during smooth sailing. people who can turn losses into fuel are the ones who truly get stronger over the long term.

source: https://x.com/kabu_kabuki/status/2062855710761959814

what these tweets together tell about japanese tech discourse this week: the conversation is skeptical of polish. local llm setups are questioned for being overcomplicated. quantum stock lists are presented as thematic gambling, not fundamental analysis. ai’s agreeableness is called out as a design flaw, not a feature. and the trader thread frames failure as the real teacher. the tone is less about breakthrough excitement and more about the gap between what tech can do and what it should do. japanese tech twitter seems tired of being sold on potential and wants tools that don’t lie, don’t fawn, and don’t require constant correction.

more at falsifylab.com

Originally published on FalsifyLab Substack.

— research and educational content. not investment, legal, or tax advice. do your own research. positions and views may change without notice.