Their strategy always was "buy company" and "instantly lay off about everyone" to save costs and rapidly increase subscription pricing (1).
So far they've been relatively soft (for their doing) on Komoot, which I too am most anxious off.
Bikepacking.com has a good read about Komoot; it was probably unsustainable in the long run before bending spoons took over anyways (2), yet I much rather had they stayed a sort of indie company driven by their passion. I will cancel my long standing Komoot subscription the day enshittification news breaks.
You can imagine all of these moderately successful SAAS companies that see peak subscribers starting to fall off on top of legacy tech stacks and no will to make drastic steps to get back to growth and understand why they sell. I've never seen BS as specifically ruining companies (although they've certainly been known to jack up prices for the remaining subscribers) but it's not a good sign when they do buy something you use.
What would you rather have? A five-year struggle to turn around a stagnant SaaS, or a big fat check? It's a simple and effective model. First one out gets the biggest check.
The "local" company is already UK owned though, so at most "European", not national or EU.
What I find strange is that the Dutch government does have its own datacenters, e.g. ODC-Noord (1), but they're still looking to outsource the hosting even after the current contract ends in 2027.
I suspect that most government departments see data centers as a liability and are very happy to outsource to the big providers, apart perhaps from the ones hosting stuff they don't really want you to know about.
It's always better to be able to blame a supplier for something going wrong if you're a senior leader or politician. For some reason, if it does happen no one has to resign.
There is loads of UK Critical National Infrastructure on AWS, probably Azure too. And the Home Office put up £10 million tender to shut down an old data centre not that long ago without a confirmed replacement - https://www.find-tender.service.gov.uk/Notice/018193-2024
Seems like governments, at least ones of the size under discussion here, are big enough to have a general services agency run a consolidated data center for the rest of the government, allowing individual agencies/departments get similar benefits of outsourcing to a commercial entity.
I've used Fastmail for years but a year ago switched to Proton.
For me the only reason to switch to Proton was that its hosted within the European continent, while Fastmail is hosted in
I would say that Fastmail is the "Ferrari of e-mail" services. It does everything well, or extremely well, especially if you have more advanced setups like wildcard domains.
In particularly, I miss being able to send from wildcard domains. While proton has a thing called simplelogin, it only works kind of seamlessly if you get an e-mail on a wildcard address and want to reply to that same address. Sending from any * domain requires you to make the address via the simplelogin page and isn't nearly as seamless. While you can make some sending addresses (i.e. regular aliases) in the protonmail interface, that's a trap, because once you've made an alias, you can't delete it unless there's no mail related to it in your mailbox anymore (even if you have a catch-all setup; I wonder if it has anything to do with how the encryption keys are setup, but it still sucks).
I also miss both snoozing and pinning mail. Officially, the proton mail apps (1) do support snoozing, but that requires "conversation view" to be enabled. I think the conversation view over groups e-mails too aggressively, and don't really understand why snoozing without conversation view isn't possible. It's utterly annoying. As far as I know, pinning e-mails isn't a thing in the proton apps. There are "stars" but these could have been labels (which also exist). They don't pin the e-mail to the top.
The proton mobile apps also lack various settings which are in the web interface, like access to sieves. The apps are sometimes a bit laggy, especially if you have a lot of e-mails, although there seem to have been some improvement on this end. I also still get double "fingerprint to unlock" requests sometimes.
Then there's theming, which I can imagine is (even) more of an opinion, but I liked the Fastmail interface more than the proton interface. I think its cleaner. Not a particular fan of any of the themes of protonmail.
I left Fastmail just as it added offline access. This was originally my biggest gripe. I might have stayed longer if they added it just before I left.
For Proton, they have been releasing a lot of new services lately. I hope they will spend a year or more, just polishing what they currently have. They did say they will spend some time on polish in a blogpost recently, but haven't really seen the fruits from this yet (or I care about different things than they do?). And I hope I will one day be able to add more domains to my account. Even with Visionary, you only get 6 domains for 6 users, and no way to add more.
I sincerely hope Proton will never add any of the AI nagging , the OP was talking about. If they do, I'll leave the instant.
I would really like to see what "appropriately licensed data" means. Cannot imagine they didn't copy all open repo's on GitHub, and can't imagine they asked for permission, or are reproducing license texts from these repo's now. It sounds hand wavy.
P.S. A fairly basic website otherwise, but it unfortunately seems to be hacking scroll for no good reason.
Presumably their position remains that training on public repos is fair use and doesn't require a license. If it doesn't require a license it's still "appropriately licensed".
I assume they took the actual repos’ licenses info account. I don’t understand why they should ask for permission when the license would already allow for it.
Almost all licenses have requirements to redistribute copies of the work, or derivatives thereof. Even permissive licenses do. It's very little to ask when open source dev's provided thousands of hours of free work.
For example, the Apache 2.0 license requires in just 4.c:
You must retain, in the Source form of any Derivative Works that You distribute, all copyright, patent, trademark, and attribution notices from the Source form of the Work, excluding those notices that do not pertain to any part of the Derivative Works;
Just because they're tokenized and transformed into a probabilistic mapping, doesn't suddenly mean that they weren't copied.
I find it morally unethical that they (likely) just ingest IP of all open source repo's without asking, but also importantly without any attribution.
Let me also note that I'm not against LLM's in general. But I do think training on open source must be opt-in, and I look forward to a world with actually ethical, and traceable (i.e. on what they were trained on, like a bill of materials (BOM)), models.
Recently, GitHub has changed their terms of service to use all user data for AI training unless users explicitly opt out. This is probably the way Microsoft has obtained "appropriately licensed data".
I've played both Factorio and Bitburner extensively (both >1000h, but that's unfair wrt bitburner, because you let it run in the background sometimes), but I don't find they compare that well. Factorio can be played without any optimization and completely mechanically if you wanted (i.e. no "programming" of circuits). It's visual style also makes interfacing with the game more like most games you encounter.
In bitburner, you literally have sort of editor (or terminal), which is also the world (you can use an external editor though). The whole game is about programming your way to destroy a BitNode.
I guess they're comparable because they are both about optimization of bottlenecks. In my experience though, having played Factorio for hundreds of hours with software engineer friends, BitBurner with its text only interface is far more niche, and only the thought of playing it reminds some of them of work (so they don't try) ;).
"AI will take over almost all the work of software engineers (SWEs) end - to - end in just 6 - 12 months!"
What you describe is >50% of the job of SWEs, even when they write all code by hand.
Are you saying that "for many start-ups", this isn't done by SWE's but by some other career type or are you implying that it's just the code written (and first review) is replaced by AI?
I have watched Dario’s interview at WEF referred to in the article and I am quite certain Dario didn’t say that. He talked about AI automating most coding already or soon, not software engineering as a whole.
He did say a few months later in an interview in India that AI will eventually take over most of SWE tasks.
—-
My statement on startups is largely about automating coding by SWEs. My startup also uses AI to automate part of technical specifications and code review but I am not sure how widespread that is.
There are other reasons why a project like Zig might not want to accept LLM generated contributions.
Zig, as programming language, has a multiplier codebase. A bug may affect a significant larger portion of users than most libraries or binaries will, as it's a fundamental building block of everything that uses Zig. Just that could be worth the extra scrutiny on every individual commit.
There's also the usual arguments: copyright ethics, environmental ethics and maintainer burden.
It might be one of the reasons they want to migrate to Rust, i.e. to handle many these memory related issues by the compiler.
Personally I used bun on a very few personal instances. But if you check issue reports, you will see memory bugs being reported say more than deno.
The Try trait (representing the ? the operation) is super cool though! I wish it was marked stable so you could implement it for types without using the nightly compiler.
Note that both Option and Result implement that same trait.
Perhaps if try blocks ever become a thing... we can finally use it for our own types ;)
The modern Try trait (try_v2) is indeed wonderful and I hope to see it stabilized one day.
AIUI a key innovation is ControlFlow, reifying the Break/ Continue choice as a sum type in the type system. This is already stable and is a useful piece of vocabulary even without its contribution to understanding the Try trait.
Knowing that Bob's CircusPerformance trait and Sarah's SeaLion type both use ControlFlow to decide whether we should keep going or halt ASAP means you don't have to write fraught adaptor code because Bob thought obviously the boolean "true" means keep going while Sarah's understanding was that it's a signal about being finished, so "true" means stop.
For Try what ControlFlow did was unlock the difference between "Success / Failure" as encoded by Result::Ok and Result::Err and the "Halt / Carry on" distinction ControlFlow::Break and ControlFlow::Continue. Often we want to stop when there's an error, but sometimes we mean the exact opposite, carry on trying things until one of them succeeds.
Finally! Glad they will now offer something which doesn't have a bending frame.
... but I wish they would make something with a bit more screen estate without being heavy and bulky. Their 16" is just too big. I really like the Dell XPS 14 and MBP 14", which I think is the right trade-off between screen size and portability.
So far they've been relatively soft (for their doing) on Komoot, which I too am most anxious off.
Bikepacking.com has a good read about Komoot; it was probably unsustainable in the long run before bending spoons took over anyways (2), yet I much rather had they stayed a sort of indie company driven by their passion. I will cancel my long standing Komoot subscription the day enshittification news breaks.
(1) https://www.dcrainmaker.com/2025/03/komoot-acquired-history-... (2) https://bikepacking.com/plog/when-we-get-komooted/
reply