mikaelaast

That’s very intriguing. I guess it would involve specifying a mental model in clear enough terms for it to be comparable to an already verified model, or something.

I like this. The few-shot example snippet method is something I’d like to incorporate in my workflow, to better align generated code with my preferences.

(Those are hardly analogous comparisons to LLM generated code, are they?)

So you do a vibe check?

If it’s in the second category, I struggle not to mentally model it. How do you stop yourself? And should you?

How do you verify the code without actually looking at it?

Creating work for the clean-up crew and leaving good money on the table for them (because it ain't gonna be cheap).

Yes, and I like the points you are making. I feel like the mental models we make are exercises in a purer form of knowledge building than the code artifacts we produce. A kind of understanding that is liberated from the confines of languages.

Yes. It would make it much easier to evaluate results if the input contents were parameterized and normalized to some agreed-upon structure.

Not to mention the advantages it would present for iteration and improvement.

Sure. Verifiability is far-fetched. But say I want to produce a statistically significant evaluation result from this – essentially testing a piece of prose. How do I go about this, short of relying on a vague LLM-as-a-judge metric? What are the parameters?

Are we sure that unrestricted free-form Markdown content is the best configuration format for this kind of thing? I know there is a YAML frontmatter component to this, but doesn't the free-form nature of the "body" part of these configuration files lead to an inevitably unverifiable process? I would like my agents to be inherently evaluable, and free-text instructions do not lend themselves easily to systematic evaluation.

I love this. I tried doing sub-pixel simulation for a tool I created (screenstab.com if anyone’s interested – yeah I know, shameless plug, etc.). I ended up abandoning the sub-pixel aspect in my shader because of the distracting patterns caused by the Moire effect.

Shameful (-less?) plug: I’ve implemented dithering for creating bead patterns from images in my web application Beadifier https://www.beadifier.com

The link again, but clickable: https://www.linkedin.com/posts/erke_chatgpt-chatgptplugins-s...

I feel like your second paragraph undermines the first one.

While Figma can be a useful tool for aligning design with code, I think it's unrealistic to expect it to accommodate all the constraints of the web platform. Relying solely on a 1-to-1 mapping between Figma components and code components can be problematic and may not accurately reflect the complexities and nuances of web development.

Acknowledging that formalization typically occurs in the development process, it's worth noting that it's often the developer who initiates and enforces it. However, it would be beneficial for designers to see formalization as an integral part of the design process itself, which could lead to more efficient collaboration between designers and developers, ultimately streamlining the entire development process.

> The designers should be working with the developers to implement their vision

I agree with the importance of this. I guess my gripe is with the fact that at the end of the day, the burden of formalizing anything that gets put on the web is on the shoulders of developers, even in the case of expressing design language, as this usually isn’t discernibly structured until the developer starts typing out code.

OP here. I propose a concept to create a tool similar to Figma but focused on designing screen reader experiences. The goal is to encourage designers to formalize and structure their work in a way that would consider accessibility and user experience for those relying on screen readers.

This new tool would require designers to think about the semantics and hierarchy of the content, forcing them to consider not just the visual presentation, but also the underlying structure that screen readers rely on. By doing so, designers would have to make their design intentions more explicit and less open to interpretation by front-end developers.

I shared this idea with my coworkers and the reception was lukewarm.

Thanks for your comment! I'm glad it caught your attention and stood out as unique. Developing this has taken an inordinate amount of effort, so it's rewarding to see it recognised. I set out to create something different, and you made me feel like I accomplished that. Thanks for your support!

OP here! I can read it fine, personally. There's a trade-off between shininess and legibility.

OP here. This really blew up. I actually made this back in 2021 (doesn't seem long ago), and probably tried posting it to HN back then, to no avail. I just posted it again on a whim, because I felt I was on a roll with my previous post on here about the metal skeuomorphism thingamajig (https://www.metalmorphism.com). If you liked this project, and want to see what I've been up to in my spare time lately, feel free to check out that discussion: https://news.ycombinator.com/item?id=34707160

OP here. LOL!

OP here. You are absolutely correct. I'm in a rural part of Norway which is nicknamed "Norges matfat" (which translates roughy to "Norway's dinner plate"), because of how so much of our produce comes from here. Moreover, I'm not exactly a company, but just one guy doing this in his free time. My day job is at a software company called Grensesnitt (check it out: https://www.grensesnitt.no)

OP here. I'm not quitting my day job anytime soon. It's a grind to get the word out in the first place, for one. The SaaS model covers some of the cost of running it, but it isn't a viable product on it's own just yet. I have other projects that generate income to cover cloud services, accounting costs and other fees.

OP here. Sorry to hear it. What browser and OS are you on?

OP here. Thank you very much for using it, and for the thoughtful comment.

OP here. Thanks! I know what I'm getting myself into when posting here. This crowd usually has strong reservations when it comes to subscription based business models, which I can understand. For me the monetization aspect is a necessary "evil" to justify spending so much of my free time outside of my day job and obligations as a family man.

OP here. As per the footer text: "Image processing is done in the browser. No image data is stored or transmitted."

OP here. I understand this instinct. In this case it doesn't actually get transferred, but stays on the client, as stated in the footer: "Image processing is done in the browser. No image data is stored or transmitted."

OP here. Sorry to disappoint you with the lack of animation. I have definitely entertained the idea of creating a video variant of this app. I fear it will remain a pipe dream due to the demands of my day job and family life.