As a Canadian / American who now lives in Europe: IMHO the two-party system and current constitutional structure in the US is an unfortunate local maximum.
It was very definitely better than the centuries of militaristic monarchic feudalism Europe waded through from medieval times until the mid-1900s. It is very definitely worse than modern pluralistic coalition-based democracies with proportional representation, which offer a wider range of choices to voters, and make it possible to launch competing parties / movements to counter institutional stagnation.
Until recently, the one counterargument I would hear to this second assertion is "but coalition governments have a hard time getting anything done". Now that we see a prime example of a government that alternates between a) not getting anything done and b) getting things done that belong somewhere in a timeframe from the 1890s to the 1940s, I no longer hear people making that counterargument.
Re: constitutional structure, one Irish friend I have made an interesting point: in his lifetime, there have been many changes and amendments to the Irish constitution. This is next to impossible in the US system, both because of the party loyalty dynamic mentioned above _and_ because of the incredibly high procedural bar to doing so. (And not least because of the current predominance of originalist thinking in the judicial branch, as though the constitution were an infallible document handed down from gods among men, eternally to be interpreted as the Founding Fathers intended back over 200 years ago in a completely different social, political, and technological context.)
"Force" is often an unrealistic expectation, though. Taking Claude Code as an example: you can add as many rules / guidelines as you want in instruction files, but they will not be followed 100% of the time, and more is not better [1].
You can of course use PreToolUse hooks to block particularly damaging actions of the "rm -rf" variety, but this is also not 100% guaranteed unless you're able to block _all_ ways of performing that damaging action (and you would be surprised: agents will happily write custom python / bash / etc. scripts to do actions you tried to block them from doing!)
Tools help instruct the agent to redo work e.g. to pass linter / formatter checks or relevant tests. But I've also seen them ignore those, often enough to be noticeable: e.g. "17 of 18 tests pass, the other 1 wasn't introduced by this feature" - regardless of whether that's actually true or not, regardless of whether I put "ALWAYS make sure ALL affected tests pass" in an instruction file somewhere.
This isn't to refute your main point: yes, you can improve your chances that AI will write good code. But there is no magic bullet that will force it, 100% of the time, to write good code; this is where vibe coders without requisite coding + engineering skills hit a wall. A multi-layered approach of guidelines + progressive disclosure + tools + hooks indeed reduces the probability of bad code enough to be useful for many engineering tasks.
As someone who now lives and works in Denmark: it's sad that so many of us have been conditioned to think 6 weeks severance is generous.
Here, labor unions are quite widespread, and very effective at negotiating reasonably but firmly. As a result, I can depend on 3 months severance _guaranteed under law_ after 6 months at a job. (After 3 years, it goes up to 4 months, and then from there up to a max of 6 months.)
It puts the responsibility for risk of instability, errors in planning hiring / capacity, etc. firmly where it belongs: with the employer.
(And no, the economic sky is not falling here as a result. Quite the opposite.)
Welcome to our cozy little country; I hope you're settling in well.
Just out of curiosity: Assuming you're a SW engineer, did you join IDA or Prosa or did you decide to not join an union? I'd like to gathers some more datapoints to help other engineers moving to Denmark make an informed decision.
> but people have hundreds and thousands on conversation on these apps that can't be easily moved elsewhere.
Except these aren't conversations in the traditional sense. Yes, there's the history of prompts and responses exchanged. But the threads don't build on each other - there's no cross-conversational memory, such as you'd have in a human relationship. Even within a conversation it's mostly stateless, sending the full context history each time as input.
So there's no real data or network effect moat - the moat is all in model quality (which is an extremely competitive race) and harness quality (same). I just don't think there's any real switching cost here.
I use OpenAI a lot on the paid plan via the UI. It now knows absolutely loads about me and seems to have a massive amount of cross conversational memory. It's really getting very close to what you'd expect from a human conversation in this regard.
Sure the model itself is still stateless, and if you use the API then what you say is true.
But they are doing so much unseen summarisation and longer context building behind the scenes in the webapp, what you see in the current conversation history is just a fraction of what is getting sent to the model.
Baffled that someone tech literate would be boasting about this in the year 2026. I mean, you do you, we all have different priorities and threat vectors, but this is the furthest from what I would personally want.
It's not boasting, I'm not sure why what I wrote would come across that way. I'm describing how I use a product and the functionality it presents to me.
But yes, it's an emerging area and I am questioning if I am sharing too much with it. I 100% would not want my chat histories exposed.
Saying that though, facebook can read my highly personal messages, google every email, my phone is tracking my every move, I have to sign up for random janky websites for my kids school where ther medical info is stored, etc.
LLM chat history presents a new risk and a different set of data, but it's a crowded minefield already.
This is the same as when Google got big (and Facebook, etc...). We have some privacy focused competitors (Kagi, etc...) but most people are quite happy to just give Google (and worse, Facebook) everything.
AI is just a new technology but this has been ramping up for decades now.
I see people who have conversations spanning months. They don't start new threads and instead go back to existing threads to continue the topic. They also reference the prior threads discussion many times.
This would feel like a switching cost for people who use the system that way.
They need to do some sort of shared chat. Like being able to start a thread then invite another chatGPT user to join on the conversation. That would add some network effects and switching cost.
Maybe they already have this? I'm not a paid user.
Is there more detailed information how this works? I used to assume that it can be beneficial to switch to a new chat to avoid having took much irrelevant context in the interaction. How does this personalization happen, how does it decide which parts are relevant from one conversation to another?
It doesn't seem like there's a way to inspect or alter what kind of information Gemini had saved as "important information" about me (apart from deleting chats entirely, apparently).
On the web app, I see the "temporary chat" option but no toggle. It tells me temporary chats aren't used for model training. I thought I remembered that chats of Pro customers aren't used for that in any case. Hard to keep track of all this stuff.
Ultimately, I think the crossover memory is useful, but I'd really like to know exactly what's in there and an ability to validate/adjust, not just on/off.
Model training is completely different than keeping a summary of chats on the side and injecting it as context.
In my Gemini app, when k click new chat and click the filters button I have “Personalize Intelligence: Personalize chat when helpful.”
It is on every time I click new chat. Maybe you need to enable it in settings first. I can disable it to have a clean chat without personal context, but preserve the chat history unlike temporary chat.
I understand they are separate processes (compacting memories vs training new models), it just surprised me to read that my chats are used for training.
This is how it's presented: "Temporary chats Opens in a new window don't appear in Recent Chats Opens in a new window or Gemini Apps Activity Opens in a new window and aren't used to train models or personalize your experience."
I'm guessing you're maybe on iOS? I don't see these UI elements, not in the App on my phone nor in direct web access.
For anyone who's a fan of the word game Wordle (where you try to guess a secret 5-letter word in 6 guesses): I've been working on a small side project I call Adversarial Wordle. Enjoy!
TBH there will likely be a _huge_ demand for "digital sovereignty consulting" over the next while, especially in the EU (and maybe also Canada).
Here in Denmark, the previously unthinkable is happening: because of Schleswig-Holstein's leadership in moving to OSS, the Danes are now seeking to learn from the Germans (or at least, that particular set of Germans) about digitalisation! That trend, plus the Danish government's all-in-on-vendors/consultants approach to digitalisation, will likely open a sizeable market - and the traditional vendors like Netcompany have taken a large beating in public opinion themselves, so it's a good time to start something in this direction.
And at the Digital Tech Summit in Copenhagen this year, digital sovereignty (and the lack thereof) was a very prominent theme across both public and private sector talks. As was the comparative advantage the EU has in _trust_, and how that helps e.g. businesses around cybersecurity, privacy-oriented SaaS, and data management expand even outside the EU - which makes it extra infuriating to see continued political interest in things like Chat Control and cracking down on GrapheneOS. This trust is IMHO pretty much the only advantage the EU has in the global tech marketplace, and we're busy throwing it away.
As a Canadian-American living in Denmark, I've seen both sides of this. In short: trust and mistrust are _both_ self-reinforcing concepts.
To take an example - would I want the current US government to be better at compiling information across all its agencies / departments? Absolutely not. What it does with its current level of consolidation is authoritarian enough that I'm not moving back there any time soon. I hear similar sentiments from my Hungarian colleagues, who are quite familiar with competitive authoritarianism in their own country.
Of course, this mistrust becomes self-reinforcing. I don't trust the US government, so I want it to be bad at its job - but then it's bad at its job, so I see it as ineffective and bloated and continue to mistrust it.
IMHO the only way out of this spiral is the hard way: a system must do the hard work to show itself trustworthy, and it must do so _before_ people will entrust it with the information that would make the job of being trustworthy easier. As with human relationships, it takes a _lot_ more work to repair trust than it does to break it. Unlike with human relationships, you also have systemic factors: the system needs an unbroken series of good, principled leaders; it needs to visibly and credibly punish corruption, not turn a blind eye; it needs to de-escalate divisions, not inflame them; it needs various institutional safeguards to work properly, not chop away at them; it needs to allow meaningful dissent and criticism, not crack down on it; it needs to learn from expertise, not undermine it.
Most importantly: the system needs to learn from its failures, and adjust the rules and incentives of the system itself to prevent those failures from recurring. This is generational work.
I see a lot of complaints about scope creep and vision drift here, so I'll add a different perspective: I use Notion right now, but have been considering a switch to Obsidian for a while now (variety of reasons, chief among them a desire to reduce my own dependence on US-based tech platforms and tools).
The lack of something like Notion databases / tables was the last thing stopping me from migrating over; I found this feature really helpful for organizing my thoughts and tasks as I want them organized, and not having it would have been a noticeable UX regression for me.
With this launch, I'll take a deeper look. It's that simple: it provides a feature many want, largely because it's seen as a killer feature of comparable closed platforms.
It was very definitely better than the centuries of militaristic monarchic feudalism Europe waded through from medieval times until the mid-1900s. It is very definitely worse than modern pluralistic coalition-based democracies with proportional representation, which offer a wider range of choices to voters, and make it possible to launch competing parties / movements to counter institutional stagnation.
Until recently, the one counterargument I would hear to this second assertion is "but coalition governments have a hard time getting anything done". Now that we see a prime example of a government that alternates between a) not getting anything done and b) getting things done that belong somewhere in a timeframe from the 1890s to the 1940s, I no longer hear people making that counterargument.
Re: constitutional structure, one Irish friend I have made an interesting point: in his lifetime, there have been many changes and amendments to the Irish constitution. This is next to impossible in the US system, both because of the party loyalty dynamic mentioned above _and_ because of the incredibly high procedural bar to doing so. (And not least because of the current predominance of originalist thinking in the judicial branch, as though the constitution were an infallible document handed down from gods among men, eternally to be interpreted as the Founding Fathers intended back over 200 years ago in a completely different social, political, and technological context.)