Hacker News: paxys

New comment by paxys in "Nine PBS sues Iron Mountain over blocked access to archival data"

paxys — Thu, 13 Aug 2026 22:19:45 +0000

All this over 50 TB of data. Could have created an in-house replicated NAS with like $2500 worth of hardware.

New comment by paxys in "How Organizations Use AI: Evidence from ChatGPT [pdf]"

paxys — Thu, 13 Aug 2026 20:42:43 +0000

Making valid complaints would require actually reading it.

New comment by paxys in "Codex in ChatGPT desktop app for Linux is now in preview"

paxys — Thu, 13 Aug 2026 20:06:59 +0000

It’s only remarkable if they actually took 6 months of engineering effort to build this. It’s more likely that it just never got prioritized.

New comment by paxys in "Accelerating GPT-5.6 Sol Ultrafast"

paxys — Thu, 13 Aug 2026 20:02:01 +0000

There is zero chance this will be offered to subscription users.

New comment by paxys in "Lovable raises $400M Series C"

paxys — Thu, 13 Aug 2026 07:41:26 +0000

They are a 3 year old startup raising a series C round. Profitability isn’t a concern, and won’t be for a long time.

New comment by paxys in "California Uber, Lyft drivers win union recognition"

paxys — Thu, 13 Aug 2026 04:20:09 +0000

The economics of self driving taxis is still to be determined. Waymo right now costs significantly more than Uber and Lyft and service is extremely limited. And Google is still pouring billions into the division with no expectation of profit in the short term. A lot has to change before self driving taxis will seriously start to displace regular ones.

New comment by paxys in "Lovable raises $400M Series C"

paxys — Thu, 13 Aug 2026 04:06:53 +0000

> How many years of insane growth would it take to pay back $400M at this rate?

Lovable reported hitting $500M in annualized revenue last month and estimates $1B+ over the next year (https://techcrunch.com/2026/06/09/lovable-says-it-has-hit-50...)

So I’m going to say not too many.

New comment by paxys in "Lovable raises $400M Series C"

paxys — Thu, 13 Aug 2026 03:41:18 +0000

They’re good enough already. You can one-shot a website with Claude Code or Codex of the same or better quality than Lovable. I have no idea what the appeal of the product is at this point.

New comment by paxys in "Grok 4.6"

paxys — Wed, 12 Aug 2026 22:44:28 +0000

The Opus 5 release was a perfect example of how useless these benchmarks are for a head to head model comparison. Anthropic published a post showing Opus 5 beating Fable in almost every eval but then added a disclaimer that it was still a tier below Fable in intelligence (and thus pricing). So then what did all the numbers represent exactly?

New comment by paxys in "Grok 4.6"

paxys — Wed, 12 Aug 2026 22:36:42 +0000

It’s equivalent to having client-side input validation. Yes it can easily be bypassed, but in the vast majority of cases where users aren’t malicious it gets the job done quickly and cheaply.

New comment by paxys in "Grok 4.6 scores 61 on the Artificial Analysis Intelligence Index"

paxys — Wed, 12 Aug 2026 18:40:23 +0000

OpenAI is almost there, and Anthropic is pretty close behind. In the next year or two all major AI companies will be vertically integrated to a good degree.

New comment by paxys in "Controversial creators are benefiting from monetization programs run by Meta"

paxys — Wed, 12 Aug 2026 15:20:53 +0000

Everyone needs their daily dose of outrage. Tell me – why are you here on HN interacting with this very post?

New comment by paxys in "Dropbox is an obvious PE Target?"

paxys — Wed, 12 Aug 2026 05:38:21 +0000

> But a lesson from this story is twofold: when Steve Jobs tries to buy you, take it.

Not quite sure how the author reached that conclusion, considering – by their own calculations – Dropbox is profitable, rich on cash flow and worth at minimum 10x what Steve Jobs offered for it.

New comment by paxys in "CFTC declares market emergency, orders Kalshi to continue to operate in New York"

paxys — Wed, 12 Aug 2026 01:04:39 +0000

Donald Trump Jr is a “strategic advisor” at Kalshi and was given equity in the company. Hope that helps.

New comment by paxys in "US hires over 2k video gamers as air traffic controllers"

paxys — Wed, 12 Aug 2026 00:36:36 +0000

Why are they saying “video gamers” as if it is a job title? Qualification for ATC is to be under 31 and the job is male dominated. No shit everyone they hire plays video games.

New comment by paxys in "Stealing Reasoning Traces from Proprietary LLM APIs"

paxys — Tue, 11 Aug 2026 23:23:15 +0000

Computer screens aren’t people either. What’s your point?

New comment by paxys in "Stealing Reasoning Traces from Proprietary LLM APIs"

paxys — Tue, 11 Aug 2026 22:20:51 +0000

They can always keep the encrypted blobs server side and send the key to the user. But regardless, that isn’t going to help with this issue (see my comment above).

New comment by paxys in "Stealing Reasoning Traces from Proprietary LLM APIs"

paxys — Tue, 11 Aug 2026 22:19:37 +0000

Encryption is irrelevant here. Even if it was kept fully server side, the actual issue is that they allow starting a conversation in a strong model and continuing it in a weaker one. Disallowing that entirely would be a huge hit to user experience.

New comment by paxys in "Stealing Reasoning Traces from Proprietary LLM APIs"

paxys — Tue, 11 Aug 2026 22:10:06 +0000

It isn’t the concern of the LLM. Regardless of where the encryption/decryption is happening, the issue is that the LLM needs to access the raw CoT.

New comment by paxys in "Stealing Reasoning Traces from Proprietary LLM APIs"

paxys — Tue, 11 Aug 2026 20:27:47 +0000

The exploit here isn’t a leaked encryption key. It’s pretty likely that they are already using a unique key per conversation. The raw CoT eventually reaches the model, and you can convince the model to share it with you.