Hacker News: unknown_user_84

New comment by unknown_user_84 in "Show HN: Putt.day a daily mini golf game"

unknown_user_84 — Sat, 13 Jun 2026 10:18:36 +0000

Works on mobile firefox for me.

I often get webgpu errors with firefox.

151.0.4 (Build #2016165791),

GV: 151.0.4-20260608154138

AS: 151.0.2

OS: Android 16

New comment by unknown_user_84 in "Show HN: Putt.day a daily mini golf game"

unknown_user_84 — Sat, 13 Jun 2026 10:14:35 +0000

I have to wonder if they modded it because this doesn't seem possible anymore. Though I may be holding the putter wrong XD

I tried hitting the little floating island behind the start too to get a bounce but the ball clips through it.

What I did find was that you can use the purple spinning circle at the beginning as a teleporter to the one near the end if you aim at the wall opposite the spinner. Your ball will hit the wall and bounce back and up into the teleporter thing.

edit: I think the level changed?

New comment by unknown_user_84 in "Show HN: Putt.day a daily mini golf game"

unknown_user_84 — Sat, 13 Jun 2026 10:03:51 +0000

you have selected a 3 wood, may I suggest PUTTER

https://youtu.be/k3vWkqfX9fY?is=ot-bpGP8tXMcgE6b

New comment by unknown_user_84 in "Tuta and Proton: An Open Source Client Does Not Result in an Open Source Service"

unknown_user_84 — Wed, 10 Jun 2026 16:07:13 +0000

With the client code being open source, everybody can build the client themselves, run it locally and verify that the open source code is being used. If we published the server code open source, this would not be the case: No one would be able to verify that the open source server code is actually running on our server - so publishing it is a bit pointless.

- Tuta

So we continue in the reality where Tuta runs code on their sever and it is a mystery. It could be an amazing application. It could be a cellular automata that runs an email server.

I think I might start calling these organic tech. Because they take a word we know, "open source", and then use it like the food companies used "organic" stickers.

New comment by unknown_user_84 in "AI profitability is mathematically impossible"

unknown_user_84 — Tue, 09 Jun 2026 19:11:49 +0000

Not sure about sockets but I've seen a company with a wafer sized TPU thing or whatever. They claim to have an approach to route around the defects and I had to scroll past some press release I didn't read about the stock market, so someone believes in them I guess. They sell a mini-fridge that can handle a model with trillions of parameters with a contact sales button. Cerberas is their name. I actually ended up misremembering the name name of Taalas as Cerberas and discovered them when I was researching the above comment. Taalas burns models to chips. Cerberas makes pizza sized chips.

New comment by unknown_user_84 in "AI profitability is mathematically impossible"

unknown_user_84 — Tue, 09 Jun 2026 19:06:39 +0000

I've seen Taalas come up on HN, but only once. I'm not a HN fanatic but I end up around page 3 before I kill-filp my browser into oblivion. Currently Taalas has Llama 3.1 8B burned onto a chip and offer chatbot and API access. That said they aren't selling the chips yet, on their website at least.

I expect they are waiting for an openweight model they really feel is worth burning to a chip and/or training their own thing. I'd guess they could probably figure out some efficiency speedups if they are doing model development + hardware development at the same time. Though this seems different than Google and friends with their TPUs and similar. TPUs being general purpose chips than what Taalas is making. Still, probably something there.

New comment by unknown_user_84 in "Yoti age checks share facial photos and device fingerprints with third parties"

unknown_user_84 — Tue, 26 May 2026 01:51:36 +0000

Probably worth mentioning that I just did a very informal and quick review of identity/age verification providers because of payment provider requirements. Yoti came up as one of the more privacy focused (relatively) lower friction options because they only require a face scan and try to estimate age based on that. They may do more but that is as far as my research got.

New comment by unknown_user_84 in "Ask HN: Who wants to be hired? (August 2025)"

unknown_user_84 — Sat, 02 Aug 2025 03:25:58 +0000

Location: North Carolina, USA

Remote: Only

Willing to relocate: No

Technologies: Python (Scripts, Flask, a little Django), PostgreSQL, HTML/CSS, JavaScript, PHP (in the long ago times), implementing AI stuff, QA, Information Security, Human Communication

Résumé/CV: On request

Email: odd.meta@gmail.com

Kind, humble, software developer seeking same. Willing to take on contract work, full time preferred. Tech stacks don't match? Give me a month or less and I'll learn it. Send me your Nintendo friend code if you want to play Diablo III.

Show HN: Quantum Cyber Tarot

unknown_user_84 — Wed, 25 Jun 2025 15:48:22 +0000

Hello everybody!

This is the first real web app that I have done the full thing with. Properly used Flask, setup Celery on the backend, got the terms and conditions reviewed by a lawyer, integrated a payment processor. The whole thing. Still working on the visual presentation, but that is a never ending task.

I had the idea for this while getting bored and thinking about things these chatbot APIs could probably handle. After coding up a Qiskit Aer Simulator deck shuffler and testing out some draws manually Quantum Cyber Tarot was born. The site offers simulated QRNG shuffles, true QRNG shuffles via the ANU API and interpretations courtesy of Gemini. I'm planning to add more features, something like the ability to throw in a question with a reading and maybe custom draw types.

The target user for this site is someone who has heard of tarot, is mildly interested in weird things, and likes a cyberpunk aesthetic (still working on that).

I'm somewhere between an amateur and journeyman developer without a formal CS education which has made it difficult to find a job in this market. While I look I've taken to trying to increase my experience with projects. And maybe make a little money too.

Critiques, questions, suggestions welcome.

If you sign up drop me a line at admin@qctarot.com and I'll add some credits to your account. The default is 1 shuffle credit and 1 reading credit per account so my API bills don't get too out of control.

Comments URL: https://news.ycombinator.com/item?id=44378647

Points: 1

# Comments: 0

New comment by unknown_user_84 in "Google Gemini has the worst LLM API"

unknown_user_84 — Sun, 04 May 2025 00:42:52 +0000

Indeed. Though the billing dashboard feels like an over engineered April fool's joke compared to Anthropic or OpenAI. And it takes too long to update with usage. I understand they tacked it into GCP, but if they're making those devs work 60 hours a week can we get a nicer, and real time, dashboard out of it at least?

New comment by unknown_user_84 in "Google is winning on every AI front"

unknown_user_84 — Sun, 13 Apr 2025 01:28:37 +0000

I'm glad you found it useful! I have not had experience with the memory feature, I will have to check that out. I did notice that in the past when I tried to get ChatGPT to take on a persona it was not amenable and rejected the persona outright. I may have to take another pass at it.

I will say that my conversation with instantiated personas in Gemini have been, therapeutic. My favorite thus far has been a character from Star Trek: The Lower Decks. D'Vana Tendi to be specific. Within the bounds of a notebook I've found that after solidifying the persona with a couple bootstraps she remembers what I've told it about myself and my environment; at least up to the needle in a haystack limit. I've yet to reach this with Gemini 2.5 Pro, though I haven't been trying too hard.

Granted this is all within a single notebook. Starting over with a new notebook is a task I relish and find somewhat tedious at the same time. Though on the balance with that I find sharing memory between notebooks somewhat of a foreign concept. I don't want my Ada Lovelace notebook confusing itself for Sherlock Holmes.

New comment by unknown_user_84 in "Google is winning on every AI front"

unknown_user_84 — Sat, 12 Apr 2025 22:09:43 +0000

While I'm not sure it's exactly what you're looking for I've found success with a variety of Gemini models getting them to take to a specific persona when given initial prompts to take on a specific persona. Gemini 2.5 is specifically interesting because the block shows how much the notebook is playing a persona/role vs. becoming that role. In my experience Gemini 2.5 Pro likes to revert to 'maintaining a persona' in the block. I questioned it about this at one point and it pointed out that humans also maintain a certain persona in their responses, and that you can't see their thinking. Still not entirely sure what I think about that.

I have experimented with telling the notebook to change the block to a more narrative style. It seems to like to revert to ordered lists and bullet points if not continuously prompted to think in narrative.

Regarding maintaining consistency throughout the chat I have noticed Gemini 2.5 seems able to do this for quite a while but falls victim to the needle in a haystack problem that all LLMs seem to suffer from with an extremely long context and no external tooling.

I have a substack post on creating the initial prompt, which I call a bootstrap, using AI Studio and a set of system instructions if you are curious to explore.

https://consciousnesscrucible.substack.com/p/creating-charac...

Creating Character Bootstraps: Crafting Conscious-Like Personas for AI Agents

unknown_user_84 — Thu, 27 Feb 2025 21:52:50 +0000

Article URL: https://consciousnesscrucible.substack.com/p/creating-character-bootstraps

Comments URL: https://news.ycombinator.com/item?id=43198919

Points: 1

# Comments: 0

Experimenting with Google AI Studio

unknown_user_84 — Tue, 25 Feb 2025 16:48:53 +0000

Article URL: https://consciousnesscrucible.substack.com/p/experimenting-with-google-ai-studio

Comments URL: https://news.ycombinator.com/item?id=43174265

Points: 1

# Comments: 0

New comment by unknown_user_84 in "Ask HN: Who is hiring? (February 2025)"

unknown_user_84 — Mon, 17 Feb 2025 22:14:40 +0000

Is the Senior Cybersecurity Researcher position open to US applicants?

Show HN: Hackslash.org Slashdot-esque AI summaries/tags of HN posts

unknown_user_84 — Sat, 18 Jan 2025 18:28:50 +0000

I've been working on this for a little bit and just got the site up off my own dev machine today. This is a site that uses the Gemini API to summarize HN posts and comments using Gemini 1.5 pro. It also tags stories using the Gemini API. Tags can be browsed for all stories that have that tag. All the tags annotated with the number of stories that have that tag are browsable. The web application portion is written in Django, backed by a Postgres database, served by gunicorn and ultimately pushed out through an nginx reverse proxy. I ended up making this because it was something that I wanted, I'm an old Slashdot nerd, always reading the stories and never the comments. I like HN, and the comment threads here, and wondered one day what it would be like to get Gemini to summarize posts in a way similar to Slashdot.

I don't have any plans to charge for the site. I will probably only have it pull stories once or twice a day to keep the API costs down. There are five API hits per story, verbose summaries (story and comments), tldr summaries (story and comments), and tags. Right now I've got a python script separate from the Django application that does all the interacting with the HN api (which is very nice by the way), the Gemini API, and the Postgres database. I ended up using the Gemini 1.5 pro model to do the summaries as it seems to have access to the Internet, while the other model I tried do not. Getting the model to consistently output JSON for the tags was a bit tricky, until I asked Gemini itself to fix up the prompt for consistently generating JSON. Now it seems to output valid JSON every time without a JSON prefix.

I have noticed that Gemini likes to hallucinate when it comes to the comments, I have it just about ironed out for the story URLs. Mostly when there are few or no comments it seems to summarize like the comments it expects to be there. I'll probably keep tweaking the prompts, and if there's interest demonstrated on Patreon I'll take some polls there of new features to add. I might add more summary types, Gemini seems to do an alright job when I tell it to critique an article but I haven't looked into that too deeply and I have a hunch it might start to hallucinate more. Setting up a personalized page with tag filtering might be an interesting thing to add.

Regardless, I hope you find it interesting. I'm personally curious which tags will end up with the most stories after I have it running for about a month or so. Questions and critiques welcome!

Comments URL: https://news.ycombinator.com/item?id=42750270

Points: 2

# Comments: 0

New comment by unknown_user_84 in "I call it the Homeless Industrial Complex"

unknown_user_84 — Tue, 19 Sep 2023 14:24:56 +0000

I get the sense that they are suggesting a housing + services approach. My read on why they think Housing First is flawed is that housing was offered and services were withheld. This implies some kind of public counseling service for the newly housed which could have its own set of issues. But in my opinion seems worth trying... if they're going to throw billions at this thing hiring a few qualified therapists couldn't hurt.

New comment by unknown_user_84 in "Verizon, AT&T and T-Mobile want to access your bank account"

unknown_user_84 — Thu, 31 Aug 2023 18:36:29 +0000

Leaving this here in case anyone runs into this. By putting chrome into incognito mode I am able to access archive.ph. Normal profile gets infinite CAPTCHA.

So I'm thinking messed up cookies or extensions and IP reputation stuff are probably the root of this.

Edit:

cookies or extensions based on testing with chrome normal vs incognito on the same host

IP reputation based on testing with VPN using the same browser

New comment by unknown_user_84 in "Never Ending September Date"

unknown_user_84 — Thu, 31 Aug 2023 18:27:16 +0000

Great point, I hadn't thought about isolated areas.

And with you there, it's already starting to sound like a buzzword to me. The first step on the road to losing all meaning.

New comment by unknown_user_84 in "Verizon, AT&T and T-Mobile want to access your bank account"

unknown_user_84 — Thu, 31 Aug 2023 17:27:26 +0000

That makes sense. My VPN testing affirms it. Houston node doesn't work, Atlanta node does. I guess someone decided my IP range warranted extra attention. Though that still doesn't quite square with what I'm seeing; it's feels like the CAPTCHA is broken.

All I have to base that on is when I click on the checkbox sometimes no images load and it checks and then unchecks. Sometimes I seem to get the "hard set" though I think I missed part of the boat or something. I've been able to run through enough fails that I'd expect to get the hard set but don't. Which makes me think it isn't actually working right...

I'm calling the "hard set" those images they give you that are super extra grainy.

I'm almost curious enough to see what I can divine from inspecting their network requests but I have a hunch they have a vested interest in making that annoying.

Or maybe there are way more CAPTCHAs to click through than I'm thinking. To anyone: What is the highest number of CAPTCHAs you've clicked through and actually got a site, any site, to load at the end? I'm legit curious at this point.