January 2026 might be the month of langs created to be used by AI. Usually the chief concern is saving on tokens, prompted by context window anxiety. (This completely disregards the fact that agents thrash the context window by doing wrong things, then attempting to fix them; or by reading unrelated stuff; or by calling unhelpful tools; etc)
Interesting take, because I think precisely the opposite. Coding agents let us produce a lot of code, code that we need to read and review. That means we need languages optimized for code generation by AI, and code review by humans.
Not a language, but we are having very good success using https://brannn.github.io/simplex/ for autonomous one-shot workflows. It seems to be a very high-fidelity input for LLMs.
Not working well with something that doesn't conform to the WICG User-Agent Client Hints specification is an interesting definition of "closed." More like, "I have standards." And it's hardly closed if you can get the information by using literally almost any other client.
Seigniorage accrues to private entities instead of the state, enriching the owners of those private entities rather than everyone in the state that issues the currency.
That’s also a downside: When your funds can be transferred away by anyone who happens to acquire the key without triggering any fraud prevention or additional verification checks, losing your entire bank account at 4AM Sunday morning becomes much easier.
Our knowledge is constantly expanding, allowing us to build things differently than we used to. Modern cryptography, which makes things like multi-sig possible, is only a few decades old; it didn't even exist when the current banking industry was being established.
In addition to the security issues, they would have to deal with non-negligible transaction costs every time they wanted to convert it to actual money so that they could purchase something. If they were using it as an investment, they had to deal with the opportunity cost of underperforming $SPY.
It really comes down to the burglar's expectations. If most crypto holders used geographically separated multi-sig, these attacks wouldn't be worth the effort anymore.
It’s the same logic as iPhones bricking themselves after being stolen. Even if your specific phone isn't an iPhone, the fact that most phones are now useless to thieves discourages the crime across the board.
This isn’t just a problem in the Netherlands or a thing of the past. 2025 actually saw the highest number of attacks ever recorded [0].
There are ways to prevent this. Like using multi-sig with geographical separation (so you can't move funds alone) or setting up forced time-delays. Ultimately, being your own bank is a massive responsibility, and I think too many people take that reality too lightly.
No, there's not. There are rarely good reasons behind what banks do because these are organizations that are run by mediocre people who are not incentivized to not suck. They don't care at all about anything. This "fraud prevention" thing only gets in my way and doesn't prevent the less sophisticated people from sending their money to India.
> If your bank doesn't want to raise the limits, there's probably good reasons behind that.
Why would their having reasons make me feel better when my payments don't go through? We complain when Apple plays nanny and makes their product a walled garden. How is a bank different? They should be doing their job without causing inconvenience for me.
It typically doesn't, it just implements compliance with laws and regulations in your jurisdiction.
Withdrawal and transaction limits are commonly such a thing, politicians get hounded because some people were frauded out of their monies and they feel a need to show that they're doing something about it.
Banking is very international, you can put your money in some other jurisdiction if you'd like to. Many transnational banks are connected to the usual payment providers, you can probably figure something out if you put your mind to it. One way to do it is to start a company, business accounts at banks generally have different limits and then you pay a lawyer or bean counter to clear how to do the books and pay appropriate taxes.
I can only really use Canadian banks because that's what Interac, Canadian system for sending money by email works with. The bank does decide what I can pay for because it limits the amount of money I can spend via Apple Pay (like $500), and in general (like $1000). You need to bring your physical card and call them to temporarily increase the limit to buy anything expensive.
Yeah all of that sounds way more convenient than your grandma irrecoverably losing her life savings because somebody kid fished her for her crypto authentication.
Web searching a bit seems to say that e.g. tourists can use their foreign issued cards and typically carry their issuers' limits with them, unless the merchant has their own limits for some reason.
Putting stricter limits on intermediate devices like pocket computers doesn't seem very tyrannical to me. Interac's web site says the typical default limit would be 2-3000 CAD, i.e. 12-1800 euros or so, unless you have a small business account, then it's more like 25000 CAD.
If you often find yourself spending thousands of CAD on a whim, perhaps it would be a good idea to open accounts with a bank that is tailored to people with a lot of money. I'm sure there are some available in Canada.
You can redeem stablecoins in blocks of a million if you are a registered bank. This is the only way to redeem them. Otherwise you can only trade them.
No, it's not. It's not possible to come to the US soldiers with a bunch of US dollars and some demands and get what's demanded in return for the dollars.
Only trust of the other market participants backs the US dollar.
Data centers in space make sense when you want it to cost 200x more than on land, be unavailable for repairs and upgrades, and be either high latency or be out of commission during periods of darkness.
It's obvious. The harder you make it to down or hijack a plane, the fewer downed planes you will see. It didn't have to be perfect to prevent and deter. Some security is better than no security. If you had no security at all you would see planes go down all the time.
And it wouldn't surprise me if some of the detection technology were classified.
It would not be "great" if governments were more open about their detection capabilities; that would cause more terrorism attempts and is one of the stupidest things one could do here.
The LLM will know how the user operates, their proclivities and brain structure, and will design UX perfectly suited to them, like a bespoke glove. They won't have to learn anything, it will be like a butler.
Google loves to reinvent shit because they didn't understand it. And to get promo. In this case, ASN.1. And protobufs are so inefficient that they drive up latency and datacenter costs, so they were a step backwards. Good job, Sanjay.
Really dismissive and ignorant take from a bystander. Back it up with your delivery that does better instead of shouting with a pitchfork for no reason.
This bystander has been using protobufs for more than ten years. I'm not sure what I need to deliver since ASN.1, Cap'n Proto and Flatbuffers are all more efficient and exist already. ASN.1 was on the scene in 1984 and was already more efficient than protobufs.
Protobuf has far better ergonomics than ASN.1. ASN.1 is an overcomplicated design-by-committee mess. Backwards compatibility in particular is much harder.
I don't doubt your experience, but with X.509 having evolved substantially, and ASN.1 on billions (if not tens of billions) of devices, in practice it seems OK. And it was formally verified early.
ASN.1 on billions of devices doesn’t make it less of an anti-ergonomic, design-by-committee piece of crap. Unless your goal is to be binary-compatible with these devices, one should be free to explore alternatives.
By all means, keep using it, but it might be worth figuring out why other people don’t. Hint: it’s not because they’re more stupid than you or are looking to get promoted by big G.
(Personally, I like the ideas and binary encoding behind Capn Proto more than all the alternatives)
One of the advantages of of protobuf I never see anyone highlight is how neat and well-designed the wireformat is, in terms of backward/forward compatibility and lowlevel stuff you can do with it. Very useful when building big and demanding systems over time.
For high performance and critical stuff, SBE is much more suitable, but it doesn't have as good of a schema evolution story as protobuf.
Yeah, like a really shitty ancient version of bash. If that's what UNIX means to you, I'm not gonna yuck your yum, but what could be more UNIX like than letting license issues make life worse for your users.
reply