I appealed but not yet, but it's only been a bit more than 14 hours. I'll try to appeal a few times if possible and try to think of other means if that doesn't help.

Varmaan ainakin @aurasalla.bsky.social miettinyt näitä asioita. Ehkä @atteharjanne.vihreat.fi myös?

Vaihtoehtoja kaksi:

1. Alustoihin (X/Twitter) ja niiden käytäntöihin vaikuttaminen
2. Säädäntö tai vähintään suositukset siitä miten yhdenmukaisesti Suomalais- ja eurooppalaispoliitikkojen pitäisi olla milläkin online-alustoilla saavutettavissa. X ei mielestäni riittävä em. riskien takia.

Mielestäni olisi syytä käydä julkista keskustelua paljonko Suomessa ja Euroopassa halutaan antaa amerikkalaisyhtiölle valtaa poliittisen keskustelun portinvartijoina.

X Has a Free Speech Problem — and I’m Living It X recently suspended my account for “violating our rules against inauthentic behaviors.” That was the entirety of the explanation. X cited…

X suspended my account for “inauthentic behaviors.” There was no warning, and no appeal answered.
When platforms become essential channels for politics, journalism, research, and public debate, opaque enforcement is not a minor inconvenience. It poses a societal risk.
medium.com/@paul-oestre...

I got suspended for "inauthentic behaviors" too, and I don't know why. Severely limits my ability to interact with Finnish politicians and participate in societal discourse. Feels absurd how much an American company can control who can and cannot interact with politicians in Finland.

Just happened to me too. Feels very arbitrary. Talked about data centers and Finnish taxation.

Suomalaisten poliitikkojen X:ään postaaminen on kyseenalaista yhdenvertaisuuden ja sananvapauden kannalta. He antavat amerikkalaisyhtiön päättää puolestaan kuka voi ja ei voi vastata heidän postauksiinsa. X antaa mielivaltaisesti banneja mm. "inauthentic behaviors" syyhyn vedoten.

My Twitter account got suspended due to "inauthentic behaviors". I have no idea what in my behavior was inauthentic. So much about free speech on X.

If it's generic people and companies can bend it to their own use cases?

I also think people have forgotten that not all the trillion dollar ideas need LLMs. There are still new Ubers and Facebooks to be invented that rely on plain old regular code.

Since in case of incident added layer of unpredictability would easily become catastrophic.

That kind of orchestration pipelines for sure would be interesting! But I think the workflows triggered should mostly be configurable rather than agentic. Something like n8n.

How would the devtools you are working on look like if you'd had 10x the resources?

I think that easy agent access is at least something that is becoming more and more crucial all the time.

Their ability to follow instructions longer is at least way better. Or at least opus 4.6.

I tried to use Mastra with claude code. The experience was horrendous: Neither the model nor me had understanding of what functions and methods to call and how so claude kept hallucinating and introducing wrong patterns. That was late last year so the situation might have became better though.

Possibly! Haven't tried making dev tools much. But i know many who have built their own agent orchestrators though.

The reason is that foundation models do not know new tech well enough. Its not in the training data, and might flow there slower if adoption becomes slower.

What I think is that it will be harder to adopt new tech. LLMs might be great woth boring tech, but working with new frameworks is very painful even with context7 and such. So brand definitely will be a moat.

I think the fear then sounds pretty generic and people on the product side have similar fears. You probably underestimate the expertise needed.

Are you afraid that dev-tooling will become obsolete in the future?

Why aren't platforms like @bsky.app and Twitter adjusting delivery of the post based on dark patterns such as vague posting? Either let the users report that they want to see less such content, or let and LLM to figure out in the background if it is a dark pattern or not.

The biggest tell for an AI post? It's not the em dash. It's the "it's not X — it's Y" sentence structure.

Myös @aurasalla.bsky.social voi tietää tästä aiheesta jotakin taustansa takia?

Näin saataisiin yhteistyö sulavaksi ja toimimaan samaan malliin miten tekkifirmat jo operoivat. Miten tälläisiä asioita pitäisi ajaa eteen päin @atteharjanne.vihreat.fi?

Google lobbaa valtioita. Miksi @hel.fi ei lobbaa Googlea?

Big tech-yrityksillä on omat government affairs -tiimit ja CSM:t. Kaupungeilla ei.

Helsingin pitäisi palkata omat CSM:t hoitamaan Googlea ja Nvidiaa yms strategisina kumppaneina ja lobbarit ajamaan investointeja.

Bluesky needs "not interested in this post" with options "not relevant" and "seen already" to avoid duplicate or near-duplicate repetitive content. Bigger problem on Twitter, but applies to @bsky.app too.

Iran Protest Death Toll Could Top 30,000: Local Officials According to two senior officials of the country’s Ministry of Health who spoke with TIME.

Up to 30 000 have died in Iran's protests per Time Magazine

time.com/7357635/more...

Trump will only act as the last reaort when Europe becomes strong enough to put stop on Russia without the US.

Theory: As long as Trump doesn't do anything about Russia, he holds the cards and Europe has to dance along. But if the US actually did something about Russia, Trump would see that as a card spent and leverage gone. He rather keeps Europe dependent.

Arkipäiväisien riippuvuuksien katkaisemiseen (sähköposti, some, streamauspalvelut jne jne) pitäisi luoda kannustimia EU:n ja valtion puolesta. Etenkin niiden kehittämiseen.