I was trying to look for the conspiracy in this but then the story started sounding really plausible

TL;Dr billionaire buys company that make UPSs that are connected directly to voting machines and allowed to have their firmware/drivers updated without checks. Same company contracts Palantir and a new Starling contract - the implication being they've snuck SIM cards into their UPSs and are communicating via Starlink from what should be airgapped systems connected directly to voting machines.

Of course why go to the hassle of Starlink when you can just use any cell provider, and surely surely hardware changes to the UPS, like adding a SIM card, are audited?

OOI what coding agent are you managing to get to work nicely with G2.5 Pro?

I mostly use Roo Code inside visual studio. The modes are awesome for managing context length within a discrete unit of work.

Scala was absolutely killing the "we need JVM because all our Hadoop era tech uses it but we're a new generation of JVM data tooling and so we need a language in which the functional programming based abstractions we're using for our data processing are first class language features" niche for a while in the late 2010s.

Was a really big part of Kafka and Spark ecosystems until they supported Python well enough that a lot of people just stuck with that instead of teaching their devs to write Scala.

Scala screwed up by having no filter. Implementing every idea someone with a PhD thought of from the entire history of programming seems like a good idea until you have to use the language. Everybody doing whatever they want because of the freedom and no guidelines didn't help. And build times killed it

Kotlin feels like it has a much better plan and it seems like so far it won't suffer the same death.

I agree about no filter*, I disagree with your reasoning. Scala (2) as a language is quite simple. The complexity came from the incredible power of its building blocks (e.g. implicits and path-dependent types). The lack of filter, as I felt, was moreso on how different two Scala codebases could be. That was part of the point though -- a scalable language, which means changing to the needs of each team.

Which also means less transferable knowledge. That killed it for me. If you muscle through it you'll get there, sure, but in the meantime I found languages that have smaller surface, are just as (or more) productive, and you can be onboarded in a project in half an afternoon.

This is true but what’s odd is that Java has better functional programming support than Python.

Could you elaborate?

A simple example is that people who start with Python think the list comprehension with all the colons and negative numbers is actually good instead of a hindrance. Python relies on pandas, Polars, and Pyfunctional to prop up a bad language design choice.

List comprehensions look like this:

  [f(x) for x in xs]

I think you mean the slice syntax (`xs[5:7]`). I don’t see how this is a bad design choice, though.

Take a look at F# for a superior list comprehension design.

In Java you can write multiline anonymous functions and in Python you can’t (well, not in a reasonable way).

Java has immutable bindings with the final keyword.

Java variables are lexically scoped, which discourages mutation of globals.

I don't understand how to use this, I keep trying to edit a photo (change a jacket to a t-shirt) of myself in the Gemini app with 2.0 flash selected and it just generated a new image that's nothing like the original

I think this is just in AI Studio. In the Gemini app I think it goes: Flash describes the image to imagen -> imagen generates a new image

It is very sensitive to your input prompts. Minor differences will result in drastic quality differences.

Remember you are paying about 4 cents an image if I'm understanding the pricing correctly.

I use Vertex because that's the one that makes enterprise security people happy about how our datas handled.

Do Google use all the AI studio traffic to train etc?

Not if you have billing enabled: https://ai.google.dev/gemini-api/docs/pricing

This is correct, "When you activate a Cloud Billing account, all use of Gemini API and Google AI Studio is a "Paid Service" with respect to how Google Uses Your Data, even when using Services that are offered free of charge, such as Google AI Studio and unpaid quota of Gemini API."

There are a few conditions that take precedence over having-billing-enabled and will cause AI Studio to train on your data. This is why I personally use Vertex

The way dependency resolution works in Java with the special, Google only, giant dynamic BOM resolver is hell on earth.

We have to write code that round robins every region on retries to get past how overloaded/poorly managed vertex is (we're not hitting our quotas) and yes that's even with retry settings on the SDK.

Read timeouts aren't configurable on the Vertex SDK.

The thing is that the entry level of provisioned throughput is so high! I just want a reliable model experience for my small Dev team using models through Vertex but I don't think there's anything I can buy there to ensure it.

OOI what's your preferred framework for that managing agent/child agents setup?

I use Roo Code. It’s very good.

I love that you're responding on HN, thanks for that! While you're here I don't suppose you can tell me when Gemini 2.5 Pro is hitting European regions on Vertex? My org forbids me from using it until then.

Yeah, not having clear time lines for new releases on the one hand, but being quick with deprecation of older models isn't a very good experience.

I feel like I need a power users walkthrough of Kagi to shows it's value, I just can't get it up for it at the moment. I barely even use Google these days and just ask GPT :-D

I mostly like that you can do your own rankings instead of a prediction system doing it for you.

I can uprank stuff like Github and Stackoverflow, and downrank stuff like eHow or W3Schools, which is quite nice.