Hacker News: simondotau

New comment by simondotau in "Show HN: Building a Markdown editor for Mac, iOS and web"

simondotau — Sun, 27 Sep 2026 21:25:33 +0000

If an app has a menu option that makes a Finder window appear showing the relevant file/folder, I would expect that to be labelled “Show in Finder”.

New comment by simondotau in "Android 17 is the first since 3.x to add new APIs without releasing to the AOSP"

simondotau — Wed, 23 Sep 2026 09:25:35 +0000

If someone says a word has a different meaning, then that necessarily affects the meaning of what they said. You are not in a position to assert that your interpretation of their words is more objective than theirs, unless you're explicitly intending to be intellectually dishonest, or come with further evidence proving their intellectual dishonesty.

New comment by simondotau in "Samsung is expected to more than double output of its HBM4 and HBM4E DRAM"

simondotau — Mon, 21 Sep 2026 03:28:20 +0000

Or we treat them like humans and make each one do job interviews to work out what tasks they’re suited for.

New comment by simondotau in "Android 17 is the first since 3.x to add new APIs without releasing to the AOSP"

simondotau — Sat, 19 Sep 2026 15:19:24 +0000

“Not up for debate” is a much scarier string of words than anything that guy has said.

New comment by simondotau in "Android 17 is the first since 3.x to add new APIs without releasing to the AOSP"

simondotau — Sat, 19 Sep 2026 15:03:42 +0000

I’m a web developer and I have never dailied Chrome or a Chromium descendant. Firefox for daily, Safari for the occasional cross-check.

New comment by simondotau in "Bonsai 2 27B: Near-Lossless Compression in a 9x Smaller Footprint"

simondotau — Fri, 18 Sep 2026 22:29:33 +0000

This isn’t analogous to repeated addition giving multiplication. You’ve switched from applying an operation nine times to counting how many times an operation can be applied.

You chose 81 because it’s the one number that makes your argument appear to work. Try the same reasoning with 80 ÷ 9.

New comment by simondotau in "Training a 4B model to produce 81% faster query plans than Postgres"

simondotau — Fri, 18 Sep 2026 15:24:43 +0000

SQL Server stored procedures don’t provide a separate tier of T-SQL functionality. Anything they do to the data can generally be expressed and executed as a T-SQL batch. If you think you need stored procedures to avoid network roundtrips in the middle of your transaction logic, you’d be wrong.

Stored procedures are essentially a crude, database-bound API layer. For serious application development, a proper service layer provides stronger contracts, authentication, testing, versioning, observability and source control in a sane general-purpose language, ideally the same one you’re already manipulating the data with elsewhere.

New comment by simondotau in "Training a 4B model to produce 81% faster query plans than Postgres"

simondotau — Fri, 18 Sep 2026 15:10:25 +0000

That’s mostly occasionally accurate.

New comment by simondotau in "Bonsai 2 27B: Near-Lossless Compression in a 9x Smaller Footprint"

simondotau — Fri, 18 Sep 2026 15:06:17 +0000

That interpretation doesn’t make sense. Repeated subtraction does not correspond to division in the way repeated addition corresponds to multiplication.

New comment by simondotau in "Bonsai 2 27B: Near-Lossless Compression in a 9x Smaller Footprint"

simondotau — Thu, 17 Sep 2026 23:33:02 +0000

"Nine times" literally means multiplied by nine, but here we're dividing by nine. It's not unintelligible (because the corrupted verbiage is so commonplace) but it is needlessly awkward. Like saying "resulted in a size reduction increase of 10 megabytes."

New comment by simondotau in "HuggingFace: Security.txt"

simondotau — Thu, 17 Sep 2026 04:05:13 +0000

https://en.wikipedia.org/wiki/Anthropic_principle

New comment by simondotau in "AWS says it can't restore some data from mideast facilities struck by Iran"

simondotau — Thu, 17 Sep 2026 03:56:36 +0000

Moral condemnation has little coercive force unless it is backed by material power.

New comment by simondotau in "Training a 4B model to produce 81% faster query plans than Postgres"

simondotau — Thu, 17 Sep 2026 03:18:36 +0000

The answer to every problem is to rebuild statistics. (And never use stored procedures. It's just a shitty API layer in the worst language imaginable, sitting outside of source control. If you need an API layer, write it in a real language, ideally the one you're already using.)

New comment by simondotau in "iOS 27, iPadOS 27, and macOS 27"

simondotau — Tue, 15 Sep 2026 13:06:49 +0000

For casual kids gaming, Apple TV and Apple Arcade is a fantastic combo. It's been plenty enough to keep my kid occupied, using hardware I already owned.

New comment by simondotau in "XCancel service is suspended until further notice"

simondotau — Tue, 15 Sep 2026 08:43:38 +0000

1 is true: if you control the web server, you control whether the content is published as readable by anonymous visitors.

2 is true: if you disagree, provide the name of that singular corporation. To be clear, you're always relying on corporations, but there's no single gatekeeper.

3 is not the same: if I see something published on "www.water.dcceew.nsw.gov.au" I can be reasonably confident that it is published by my local Government. If I see an X account of @NSWDCCEEW_Water then I need to start looking for context clues, most of which can be faked.

New comment by simondotau in "XCancel service is suspended until further notice"

simondotau — Tue, 15 Sep 2026 08:20:28 +0000

1. It's always publicly readable;

2. It is not reliant on any singular corporation, or trusting any singular corporation.

3. The established domain name system proves authenticity.

And of course the information can still be posted to X, etc.

New comment by simondotau in "XCancel service is suspended until further notice"

simondotau — Tue, 15 Sep 2026 08:11:57 +0000

That's what some motivated cable news researchers decided to abuse statistics to say. To be generous, I would say the article makes a significantly stronger causal claim than the experiment supports. The best description is a good exploratory algorithm audit with weak statistical and causal inference.

The analysis rests on just nine synthetic accounts, treats thousands of highly correlated feed impressions as if they provide robust independent evidence, lacks a proper baseline for the political content available on X, and reports no meaningful uncertainty, significance testing or adequately powered replication.

New comment by simondotau in "XCancel service is suspended until further notice"

simondotau — Tue, 15 Sep 2026 08:06:50 +0000

You substituted a different question ("who owns the platform") for the one I originally raised ("how its information and narratives are governed in practice"). Call that a red herring or simply a non sequitur; the label isn't important. It didn't rebut the argument.

You're missing the point of those examples. Of course you can reverse them -- and that is precisely my point! Organisational form or ownership tells you very little, by itself, about whether an institution behaves well. My criticism of Wikipedia concerned how authority over contested information emerges and is exercised. "But Wikipedia isn't owned by an evil oligarch" doesn't answer that criticism.

New comment by simondotau in "XCancel service is suspended until further notice"

simondotau — Tue, 15 Sep 2026 08:00:23 +0000

The key difference is that Reddit has never been treated like a synonym for objectivity.

New comment by simondotau in "XCancel service is suspended until further notice"

simondotau — Tue, 15 Sep 2026 07:57:16 +0000

In terms of species-level harm arising from the ways their systems can be abused, I’d rank them roughly as follows. To be clear, I’m ranking a platform’s potential for species-level harm based on its scale and its ability to concentrate, enforce, or algorithmically amplify a shared narrative or behaviour while limiting meaningful challenge or escape.

  S-tier: Instagram, TikTok, YouTube [shorts, recommendations]
  A-tier: Facebook
  B-tier: Wikipedia, Reddit
  C-tier: X, YouTube [subscriptions]
  D-tier: Most major LLMs, e.g. ChatGPT.
  Null-tier: Bluesky, Truth Social, Mastodon, etc. (too irrelevant to have any meaningful species-level effect.)

I can imagine how this analysis might look bonkers at first blush. But I strongly account for how much disagreement is surfaced. Say what you will about X, at least the arguments are visible for all to see. By contrast, the proportion of Wikipedia readers who also read the talk pages to understand active disputes probably rounds to zero.

LLMs such as ChatGPT are problematic because they can reflect and reinforce prevailing views embedded across their source material, most significantly Wikipedia. But their conversational nature provides an important escape hatch: you can challenge an answer, interrogate its assumptions, ask for competing interpretations, and explore multiple perspectives.