I would wager it is being used for such things, but those users have managed to mask their activities. Give it a little more time, and we will find the bad actors that have been using them recklessly since their release.
Anonymous records feel like one of those features that are obviously useful once you work with JSON-heavy systems, but surprisingly uncommon in statically typed languages
If a statically typed language has no support for record type (or the support it does have is insufficient for your purposes) then you are forced to find other ways to express your logic, so over time those other ways start feeling more natural due to familiarity, and then you end up feeling that you don't need record types (or more extensive support for them), which in turn leads language designers/implementors to leave it out, and so the cycle repeats.
I'm not sure I agree that anonymous records are “obviously useful” when working with JSON. JSON is not typed, so if you want your deserialized JSON to be statically typed, you need to declare typed records outside of the JSON, so I'd say it's far more useful to use type declarations that are not anonymous.
That said, anonymous types are incredibly useful in general. I work with C# a lot and its support for them is rather shallow. In particular, there is no “with” expression equivalent.
The original paper that kicked off the movement a few years ago used Gaussian blobs because gaussians are differentiable and the tech to analyze a scene requires the tech used for rendering to be differentiable.
Yes, that's exactly its main purpose!
I initially started because I needed a dataset browser. I work with clinical trials, so we usually get raw data files in all possible formats, from CSV to EXCEL and, of course, SAS formats.
But since I was already using DuckDB, I thought about extending it a bit further, so you can quickly get a glance at the data.
I live by the sea, it was 28C last night, will be 30C tonight, I absolutely need AC. Sadly I have to get a mobile AC unit, which are the worst, but well :/
Unfortunately, model quality is not the only criterion for users, and often not even the most important one. Adoption is also driven by marketing, UX, integrations, pricing, ecosystem, and a lot of other non-benchmark factors.
Also, model providers are not interested to have their models compared head-to-head under identical conditions. And “Model A is better than Model B” is almost meaningless by itself. Better for what task? With what prompt? What inputs? What budget? What failure tolerance?
It would be nice to have a place where users could run their own benchmarks, define evaluation criteria for their actual use cases, and make those runs verifiable by others.
Totally agree. I used to work with a team that built a project for creating ontologies of Git repositories. The goal was to help LLMs onboard faster and navigate the repo better.
In the end, it became heavy overengineering: people no longer understood not only the repo itself, but also the extra layer describing it. Meanwhile, coding assistants are already quite good at reading codebases directly.
It would be nice to see some metrics. I think the missing layer here is evaluation. If agents are going to produce applications, the platform needs not only guardrails, but public-ish evidence that those guardrails actually catch failures
