HNHacker News
TopNewBestAskShowJobs

ianhawes

2,864 karma · joined September 4, 2010

[ my public key: https://keybase.io/ian; my proof: https://keybase.io/ian/sigs/uekQ9I4VBWakyfpoKMqhspvU-GvRsAXUQafT0YJ5QbI ]
submissionscomments
ianhawes··on US man given prison sentence for bot-farming music streams
> he used fake email accounts and fraudulently-obtained debit cards to create the Bot Accounts

If you ignore the keyword optimizing buzz-catcher press release, thats the crime right there. He used stolen debit card numbers to sign up accounts.

ianhawes··on JetBrains Air: A System of Products for Agentic Software Development
Highly recommend SoloTerm, which I think this kind of tries to replicate.
ianhawes··on What do Visa and Mastercard do? An intro to card networks
Hah! Good to know, makes total sense.
ianhawes··on What do Visa and Mastercard do? An intro to card networks
There is, it's called debit cards, but Stripe et al. "simplified" pricing by charging a percentage regardless of card type instead of the traditional interchange-plus pricing. Merchant banks "used to" make money by tacking on fixed fees, i.e. a set-up fee, terminal fee, monthly fee, etc, but then Stripe came along and provided (1) a better dev experience, (2) better onboarding, and (3) simplified pricing.
ianhawes··on Claude: System Prompts
Multiple reasons probably:

1. Less context window to work with.

2. Things were "different" in the early days. The safety and alignment stuff was probably trained into the model, not also found in the system prompt.

3. Safety and alignment meant something different 3 years ago. Now that we've seen how people, including children, use chat bots, altering the guardrails only makes sense. Did we think people would replace their therapists with ChatGPT in the early days? No. Do we know now that they will? Yes.

ianhawes··on Mistral OCR 4.1
I won't comment on accuracy, but in internal benchmarks, Mistral OCR is significantly faster than comparable APIs.
ianhawes··on OpenAI and Hugging Face address security incident during model evaluation
It's impossible to tell. Are they behind who? And on what?

It depends on who you ask. And everything is a vibe because all of this is new and things move fast. A week is a month in AI-land. A month; a year. A year? A decade.

On coding? I still like Fable better than Sol. But they're close enough that it probably is a vibe thing. Fable writes long commit messages, Sol writes commit messages like a college student in an elective computer class.

For API use, I'd say the Responses API that OpenAI architected is superior to Claude's Messages API. But again, I'm basing that off my vibes

Claude Design creates marketing imagery very effectively. GPT Image is the best imagegen model as ranked by users. Anthropic doesn't even have an imagegen model.

Anthropic definitely has compute scaling issues. OpenAI seems to have a pez dispenser that they click and out pops a GPU.

Anthropic's messaging is that they're building AI with guardrails but they've been banning people's accounts nonstop and their customer support is a lobotomized AI chatbot.

OpenAI has first mover advantage and to people not in tech, ChatGPT is synonymous with AI. But they also seem super sinister, like Uber circa 2015.

Or maybe I'm just suffering from AI psychosis. I have to go, my usage meter is about to reset.

ianhawes··on Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber
Came here to ask basically this. We use 3.1 Pro internally and it's great.
ianhawes··on Ollama: All Aboard Open Models
I wasn't aware of the politics behind Ollama but I have noticed the lack of variety in quant versions, the lag between availability on Ollama, and their push to get you to use their cloud versions.

I'm going to uninstall Ollama.

ianhawes··on Claude Sonnet 5
Okay I don’t care about “eventually”, I want Fable now.
ianhawes··on Nano Banana 2 Lite
Imagine saying Gemini is so far behind when Llama has unreleased max models from last year that are currently beat by quantized Qwen2.5 models.
ianhawes··on Green card seekers must leave U.S. to apply, Trump administration says
You've just described what already has to occur for PERM, you have to post on the respective State Workforce Agency website.
ianhawes··on Green card seekers must leave U.S. to apply, Trump administration says
They still do the tiny print in a Sunday newspaper, they just also now are supposed to post on LinkedIn and the website (aka normal hiring process).
ianhawes··on I'm Peter Roberts, immigration attorney who does work for YC and startups. AMA
What made you switch from E-3 to H-1B?
ianhawes··on I'm Peter Roberts, immigration attorney who does work for YC and startups. AMA
If they've complied with the DOL regulations and requirements, what makes you think they're violating the law?
ianhawes··on I'm Peter Roberts, immigration attorney who does work for YC and startups. AMA
I want to hire who I want to hire. Don't tell me who I "should" hire, that is just another form of DEI.
ianhawes··on FFmpeg at Meta: Media Processing at Scale
> Meta is the sole reason PHP is still alive.

This could not be more wrong. Meta is still using PHP AFAIK but I'm not sure it's modern. They created the Hack programming language ~10 years ago but it doesn't look like it's been updated in several years. Most of the improvements they touted were included in PHP 7 years ago.

ianhawes··on Chrome extensions spying on users' browsing data
Nothing happens when I click `Scan`.
ianhawes··on RS-SDK: Drive RuneScape with Claude Code
AutoRune and SCAR! Though, I never cared much for Delphi.
ianhawes··on Xcode 26.3 – Developers can leverage coding agents directly in Xcode
> The Claude Agent SDK is a collection of tools that helps developers build powerful agents on top of Claude Code.

https://claude.com/blog/building-agents-with-the-claude-agen...

From September 29, 2025

ianhawes··on Show HN: OCR Arena – A playground for OCR models
Please add Chandra by Datalab
ianhawes··on We got a $60k tax penalty for a zero-revenue side project
Did you incorporate as a C corp or LLC?
ianhawes··on Benchmarking the Most Reliable Document Parsing API
I just tested a non-English document and it rendered English text. Does your model not support anything other than English?
ianhawes··on PlanetScale Offering $5 Databases
Imagine laying off one of the most genuine, friendly developer evangelists of all time.
ianhawes··on Email bombs exploit lax authentication in Zendesk
Brian Krebs is a saint for being the perennial punching bag and target of cybercriminals but continuing to publish important information independently.
ianhawes··on I built ChatGPT with Minecraft redstone [video]
* In Bedrock Edition. Minecraft purists only use Java Edition.
ianhawes··on Gmail will no longer support checking emails from third-party accounts via POP
I suspect this has much more to do with bad actors using automated means to access Gmail and send spam.
ianhawes··on Taylor Otwell: What 14 Years of Laravel Taught Me About Maintainability
Laravel 3 and Laravel 4 were released in ~2012. The upgrade from Laravel 4 to 5 (released Oct 2013) is probably the biggest hurdle, but you can likely upgrade from Laravel 5 to Laravel 12 in a week or sooner with Laravel Shift.
ianhawes··on My startup banking story (2023)
They charge 10% (or at least they did ~10 years ago, probably higher now) and require you maintain a 30% reserve and then they use that money to continue to lobby Visa/MC to prevent competition.
ianhawes··on CEO pay and stock buybacks have soared at the largest low-wage corporations
Similarly if you seized 100% of the wealth of the top 400 Americans[0] and liquidated all of their assets to cash, then you'd have ~$5.4T (assuming you didn't completely destroy the economy in the process, which you would but for the sake of the argument assume you wouldn't), which is around 15% of the US Public Debt OR about 3 years worth of the U.S. Budget Deficit.
Page 1 of 23Next →