Hacker News: pseudosavant

New comment by pseudosavant in "Cursor launches Origin, GitHub alternative"

pseudosavant — Tue, 18 Aug 2026 17:27:35 +0000

Not in the same way really. There aren't examples of OpenAI, Anthropic, or GitHub/Microsoft doing the sketchy things that xAI has been involved with. Those companies care a lot about the reputational damage that would come from flouting their customer's data privacy obligations.

Cursor/xAI was making a full copy of every repo it ever touched within the last month, and when caught, just acted like that was an accident. A system to download every repo you touch, and then upload it to persistent cloud storage, doesn't just happen by accident.

New comment by pseudosavant in "Cursor launches Origin, GitHub alternative"

pseudosavant — Mon, 17 Aug 2026 21:45:31 +0000

What kind of enforceable guarantees could we actually get that xAI won't use our private code for training? Not empty promises, guarantees that have legal teeth. This applies to anything with xAI or Cursor though, especially after last months' revelation that Cursor was uploading every repo it touched.

New comment by pseudosavant in "Grok 4.6"

pseudosavant — Thu, 13 Aug 2026 06:44:00 +0000

https://news.ycombinator.com/item?id=49105411

Nice semi-colon. Written by AI?

New comment by pseudosavant in "HTML over WebSockets: real-time SPAs with barely any JavaScript"

pseudosavant — Wed, 12 Aug 2026 17:49:35 +0000

Definitely a well reasoned counter point to choosing websockets over SSE.

New comment by pseudosavant in "Nvidia's Risky Business"

pseudosavant — Tue, 11 Aug 2026 18:31:59 +0000

It could be quite a while before we reach that point though. 5+ years easily.

I've been keenly interested in the ability to run local models, but the hardware is just not there. Consumer RAM speeds and capacity will have to significantly increase before local models will be able to perform as well as even the lowest end GPT-5.6 Luna model.

This is on the backdrop of RAM becoming prohibitively expensive. And without the speed and quantity of RAM, it becomes impossible to generate tokens at interactive speeds, regardless of model. There is a fundamental dependency between calculating all of the active params with the given RAM speed.

Even with a model that has been quantized all the way down to Q4, the DGX/RTX Spark chip with 128GB of RAM can only generate ~18 tokens/sec for a MoE model with only 30B active parameters. There haven't been any broadly useful models below 30B active parameters. And that is for a $5000+ piece of hardware that will be one of the best for running on-device models.

I really want to buy instead of rent my AI, but the economics are truly terrible.

New comment by pseudosavant in "Amazon Is Creating the Biggest Pollution Source in the Country"

pseudosavant — Sat, 08 Aug 2026 22:04:08 +0000

These lucky Texans get to enjoy all this AI data center natural gas powerplant pollution.

New comment by pseudosavant in "Can Intel finally beat ARM on performance per Watt?"

pseudosavant — Sat, 08 Aug 2026 21:32:44 +0000

Of course, that same "phone" processor gets to operate with battery and thermal headroom of a laptop in the Macbook Neo. The bigger trade-offs are due to the device the chip operates within.

New comment by pseudosavant in "Can Intel finally beat ARM on performance per Watt?"

pseudosavant — Sat, 08 Aug 2026 21:29:32 +0000

Agreed, losing the headphone jack on something the size of a phone was already questionable space saving for a bigger battery. But on something the size of a laptop, it is a joke. I'm sure it has more to do with saving $1 per laptop than any user benefit.

New comment by pseudosavant in "xAI, SpaceX, and the Race for AI Buildout"

pseudosavant — Thu, 06 Aug 2026 20:23:36 +0000

It's been flagged now. Upset too many Elon bots and fanboys.

New comment by pseudosavant in "xAI, SpaceX, and the Race for AI Buildout"

pseudosavant — Thu, 06 Aug 2026 20:22:37 +0000

Hard to call them "irrelevant details" when they are facts, and that this data center is specifically used to run this AI for anti-semitic, pro-Hitler, and CSAM content. It would be bad enough if the data center was used to research cures to cancer, but instead it is creating nudes of children.

New comment by pseudosavant in "Video2NAND – Abusing video codecs for great computational power"

pseudosavant — Wed, 05 Aug 2026 00:17:20 +0000

Fair enough. Maybe not that far, but perhaps do a little one thing with these NAND gates they theorized? My issue may even be more with the HN title than the blog post, because the post never suggests anything like using a video codec for "great computational power".

New comment by pseudosavant in "Xbox goes down. You can't play games you own on disc"

pseudosavant — Tue, 04 Aug 2026 23:32:01 +0000

I guess I view 75% compatibility of the most oddball console architecture ever to be fairly successful. And it is still meaningfully gaining. It was 70% at the start of the year.

New comment by pseudosavant in "Video2NAND – Abusing video codecs for great computational power"

pseudosavant — Tue, 04 Aug 2026 23:23:15 +0000

Cool that you can make a NAND gate like this, but I was more hoping for something that showed how to practically (ab)use a video encoder for "great computational power" to do something real. Not just how to make a NAND gate with one.

New comment by pseudosavant in "Xbox goes down. You can't play games you own on disc"

pseudosavant — Tue, 04 Aug 2026 23:13:53 +0000

RPCS3 has come a long way in the last few of years, both in performance and compatibility. Even Metal Gear Solid 4 is supposed to be playable (albeit using game patches) now. And emulating the PS3 architecture, with those weird Cell SPUs, has to be one of the most difficult emulations to implement.

I'm continually impressed by the emulation scene's passion for maintaining access to older games. LLMs/agents will be a force multiplier for that kind of skill and passion.

New comment by pseudosavant in "Xbox goes down. You can't play games you own on disc"

pseudosavant — Tue, 04 Aug 2026 19:07:00 +0000

For games that require online services, this will always be an inherent issue. But for games that don't, I'm betting on emulation continuing to save the day. When the PS3 was released with its complex/novel architecture, it seemed like an emulator was highly unlikely, but look at how good RPCS3 is now.

On the Microsoft side, they've been incredibly good about first-party backwards compatibility. I bought most of my Xbox 360 games digitally, and I have been able to play them on my Xbox One, Xbox Series X, and now they are even bringing them to PC. My son and I were just playing the timeless classic NBA Jam: On Fire Edition (a 15yo game) a couple of days ago.

Emulation development will probably be significantly enhanced by the ability to use LLMs as well.

New comment by pseudosavant in "MariaDB: Promote getting to 10k GitHub stars in server log and client prompt"

pseudosavant — Tue, 04 Aug 2026 18:59:01 +0000

Merged 6 months ago. This hardly seems newsworthy, even if it was at the time it was being merged. `enshittification` being thrown around for a single console line on a piece of software you probably aren't paying for seems like a supremely entitled free loading response.

New comment by pseudosavant in "Long Range Wi-Fi – Pushing 2.4 GHz Wi-Fi to the limits (2019)"

pseudosavant — Sat, 01 Aug 2026 07:29:40 +0000

I've been running a 1.1km 5Ghz 802.11ac wifi link between two highly directional antennas for three years now. They have line-of-sight, albeit just barely. The kit I got cost $100 at the time. It was quite easy to setup even and supposedly can go much further than I'm using them for.

https://www.amazon.com/dp/B083JRFQ24

New comment by pseudosavant in "13 Models and 4 Agents on SWE Tasks: Go, Java, Python, Rust, TS"

pseudosavant — Fri, 31 Jul 2026 22:48:45 +0000

Distillation. It isn’t just Chinese models doing that to OpenAI and Anthropic.

New comment by pseudosavant in "Anthropic AI Models Hacked Three Companies During Tests"

pseudosavant — Fri, 31 Jul 2026 02:23:29 +0000

This seems like desperate headline attention seeking. "Oh... you saw OpenAI accidentally hacked a company? Well, we did it too! In fact, we did it 3 different times! #winning"

New comment by pseudosavant in "Kimi K3-256k"

pseudosavant — Thu, 30 Jul 2026 22:24:13 +0000

In practice, how have you used this? Are there scenarios where you need to be deploying .NET but want the ergonomics to be more like Python?