Max Woolf

Many recommend setting up an online portfolio for these types of positions. But I've applied to a number of Data Analyst/Scientist jobs recently and I am immediately rejected almost every time despite highlighting my blog/portfolio (http://minimaxir.com) and my GitHub with open-source code/Notebooks for each and every post (https://github.com/minimaxir), both of which have topped HN on occasion. Internal recruiters have hinted that my Software QA Engineer background + no CS degree implies I have no technical skill.

>> - Put a strawberry in the left eye socket. >>- Put a blackberry in the right eye socket. >> All five of the edits are implemented correctly This is a GREAT example of the (not so) subtle mistakes AI will make in image generation, or code creation, or your future knee surgery. The model placed the specified items in the eye sockets based on the viewers left/right; when we talk relative in this scenario we usually (always?) mean from the perspective of the target or "owner". Doctors make this mistake too (they typically mark the correct side with a sharpie while the patient is still alert) but I'd be more concerned if we're "outsourcing" decision making without adequate oversight. https://minimaxir.com/2025/11/nano-banana-prompts/#hello-nano-banana

I like the hand-holding because it's a better test of how well models can follow more detailed instructions. I was inspired by Max Woolf's nano banana test prompts: https://minimaxir.com/2025/11/nano-banana-prompts/

Awesome visualisation, and great write-up. On mobile (in portrait), a lot of longer titles get culled as their origin scrolls off, with half of it still off the other side of the screen - wonder if it'd be worth keeping on rendering them until the entire text field is off screen (especially since you've already got a bounding box for them). I stumbled upon [1] using it that reflects your comments on comment sentiment. This also reminded me of [2] (for which the site itself had rotted away, incidentally) - analysing HN users' similarity by writing style. [1] https://minimaxir.com/2014/10/hn-comments-about-comments/ [2] https://news.ycombinator.com/item?id=33755016

I was interested to see this on the heels of the discussion about tipping/threatening LLMs a few days ago [1] (discussed here: [2]). Based on the examples at the end of the paper, it looks like their inquiry was somewhat orthogonal to the tip/threat investigation. That is, the tips and threats can be done in a way that is polite, when "rude" is defined as calling the LLM a scumbag. Putting these two studies together, I wonder if the best outcome could be achieved by politely asking the LLM to do XYZ, offering a tip if it succeeds, and noting that terrible things will happen if it fails. 1: https://minimaxir.com/2024/02/chatgpt-tips-analysis/ 2: https://news.ycombinator.com/item?id=39495476

This is a great (pre-GPT-4) post that goes into the system prompt and its effects: https://minimaxir.com/2023/03/new-chatgpt-overlord/ Are you using any system prompt in your API calls? Note that it does have an explicit domain separation from the user and assistant roles. I suspect it plays a greater role in GPT-4 as well, i.e. was trained on a wider variety of system prompts to guide the output of the model.

I would love a breakdown of how they made the incredible header image (https://minimaxir.com/2023/03/new-chatgpt-overlord/featured.png)

just sharing this great minimaxir article countering the hype: https://minimaxir.com/2020/07/gpt3-expectations/

Google indeed can be your friend here. One analysis I like is from 'minimaxir: http://minimaxir.com/2014/02/hacking-hacker-news/ There's also a Quora answer: https://www.quora.com/When-is-the-best-time-to-post-on-Hacker-News-to-… And wouldn't you know it, a submission on Hacker News itself for a post entitled "Best Time to Post? It’s Irrelevant": https://news.ycombinator.com/item?id=9426040 I'm sure there are others as well.

I'm excited to see @minimaxir (http://minimaxir.com/) comes up with based on this data.

I am going to follow this advice! Here is a sample that I am following. http://minimaxir.com/portfolio/

thx @dd267, not bad by the way, this is an amazing post. Did this post by @minimaxir inspire this work? ~ http://minimaxir.com/2014/10/hn-comments-about-comments/

Does anyone know what "R tricks" was used to generate the graph depicting when the best time to post? I quite like that visualization and would like to use it for something else. "2.5 seconds, 2.39 GB processed, and a few R tricks results in this:" http://minimaxir.com/img/reddit-bigquery/reddit-bigquery-2.png

It's not my article. And I agree with you on that point, where I believe this article has value is in this quote: "You have to produce content that someone else would willingly share without being coerced." As I try to think of ways to make it to #1 on HN, a concern of mine has been not submitting too much "flop" content and "spamming" the system. His suggestion is to allow someone else to submit it, and that validates that there is value to at least someone other than just promoting yourself. That's a valuable thought and reminder. At least is was to me. :p Follow up: I remember your http://minimaxir.com/2014/07/show-hn/ post. Pretty interesting stuff.

The problem is you might not be aware you're making a shady feature. As an example, if you are making an ad for your app fooer. You A/B test some banners on an ad distribution network. You find that a simpler ad with a "download now" hyperlink graphic proves effective. No big deal right? People are going to be seeing these banners on places like news sites right? It's not tricking people. It's also pretty reasonable right? People are used to clicking on hyperlinks, it overcomes the resistance to click on images. A reasonable move for a banner ad to make. Except the ad distribution network serves content sites like softpedia as well. You saw the spike because you emulated the download button right under the mirror list, and people were confused. They downloaded your application under the pretext of getting something else. Or what about if I want a "Hey you should sign up for notifications for other job offerings in this field" message box? I'd love to have one of those in my application. I A/B test to see that this message box drastically increases signups. Except it was because people thought it was a paywall. The road to hell is paved with good intentions. Consider this case about the Weebly blogging service: http://minimaxir.com/2013/05/overly-attached-startup/ They were A/B testing messaging for first week signups. This hyper aggressive and insane route was actually considered. Do you think they're black hat? I don't think so, but what if this had become standard…

Correct. According to https://minimaxir.com/2026/05/openrouter-hy3/#llm-economics-in-2026:~:… [0] when served by DeepSeek, Cache Read Costs/Input Costs are a very low percentage: DeepSeek V4 Pro 0.83% DeepSeek V4 Flash 2% Notice that OpenRouter response caching is not available when account-level ZDR is enforced [1] [0] https://news.ycombinator.com/item?id=48317294#48317823 [1] https://openrouter.ai/docs/guides/features/response-caching#limitation…

Interesting, Haiku results seem to be consistent with https://minimaxir.com/2025/10/claude-haiku-jailbreak/ Author tried tried progressively harder jailbreaks against against the major models. Haiku 4.5 not only refused but got genuinely annoyed about the attempts, like it took the jailbreak personally unlike the other models. Interesting to see that same pattern show up here

Does the refrigerator magnet system prompt leak [1] still work? [1] https://minimaxir.com/2025/11/nano-banana-prompts/#hello-nano-banana:~…

That's less and less true https://minimaxir.com/2025/11/nano-banana-prompts/

Will update the post to capture that interesting tidbit, thanks And also, thanks for this post! https://minimaxir.com/2025/02/embeddings-parquet/

This may be an issue with default settings: "Modern LLMs now use a default temperature of 1.0, and I theorize that higher value is accentuating LLM hallucination issues where the text outputs are internally consistent but factually wrong." [0] 0 - https://minimaxir.com/2025/05/llm-use/

> This is because all systems can be improved upon recursively Until it becomes cosmic code. ( https://minimaxir.com/2025/01/write-better-code/ / https://news.ycombinator.com/item?id=42584400 )

Have a look at this analysis by @minimaxir about offering tips to LLMs: https://minimaxir.com/2024/02/chatgpt-tips-analysis/

Remember to tip your LLM https://minimaxir.com/2024/02/chatgpt-tips-analysis/

> People are out there claiming that GPT is doing all their coding for them. I just don't see how, unless they simply did not know how to program at all. I doubt it and certainly not for anything beyond basic. I've seen (and tried GPT's for code input a lot) and often they come back with errors or weird implementations. I made one request yesterday for a linear regression function (yes, because I was being lazy). So was chatGPT... It spat out a trashy broken function that wasn't even remotely close to working - more along the lines of pseudo code. I complained saying "WTH, that doesn't even work" and it said "my apologies" and spits out a perfectly working accurate function! Go figure. Others have turned to testing tips or threats, which is an interesting avenue: https://minimaxir.com/2024/02/chatgpt-tips-analysis/

Not sure if you've seen, but the person you are replying to wrote a blog post recently on their experiences of "getting stable output": https://minimaxir.com/2022/09/stable-diffusion-ugly-sonic/

tips: do prompt engineering like you're programming VQGAN GAN image generators can be tricked into giving results with phrases like 'unreal engine', 'artstation', 'junjo ito', 'van gogh' https://minimaxir.com/2021/08/vqgan-clip/ same for google. 'gist' gives you a specific category of technical results. seems like the default is some combination of medium, SO + github issues

Imgur only exists because Reddit at the time didn't have native image host but since they introduced it Imgur is in decline[0]. [0] https://minimaxir.com/2017/06/imgur-decline/

This piece my minimaxir is a nice (less-bloated) article about GPT-3 that can help you see few more things like the limitations https://minimaxir.com/2020/07/gpt3-expectations/

So I was researching on product launches. I'm surprised this hasn't been submitted before. Here is a post summarize the research: https://thenextweb.com/insider/2018/01/19/much-needed-research-shows-e… I also found a similar post on Reddit: http://minimaxir.com/2015/10/reddit-bigquery/ Any other posts like this for HN or other communities that you know of?