Show HN: Hacker News frontpage as a print newspaper that you can personalize

This is pretty cool, thanks for sharing!

It's reminiscent in some ways of Slashdot of yore, which would include a slug describing the submission. One of my more persistent issues with HN is that the 80-character-limit title gives parlous thin information on whether or not a submission is worth reading. Additional microcontent, even just another 120-240 characters (10--20 words or so) often helps greatly, and your project demonstrates this.

Auto-selecting slugs is of course itself somewhat fraught, one example on the front page of YourHackerNews as I write has a slug beginning "This website is using a security service to protect itself from online attacks. The action you just performed triggered the security solution...", which is probably not what you'd prefer: <https://www.tokyodev.com/articles/the-english-paradox-four-d...>.

I'm not a fan of animations as noted in another comment. The "Passport Photos" story has a hero image which animates: <https://maxsiedentopf.com/passport-photos/>. One option would be to permit removing an item entirely from the layout. Hitting "X" on a story does not presently do this. HN itself has "hide" feature accomplishing this.

In general, I would strongly caution against auto-including images from sites, particularly as those can be pathways to future abuse, including the appropriation of the image-hosting site by unsavoury content. I'd run across an example from an earlier HN submission a few months ago, of the ever-more-aptly named "ShadyURL": <https://news.ycombinator.com/item?id=41002786>.

On layout: Traditional print newspapers aren't merely an assortment of articles, or a ranked placement of articles, but an edited presentation of them. There's usually a top story, of course, but gathered around those will be stories related to the primary feature. See a recent archive of the (online) NY Times homepage for an example: <https://archive.is/HO7xW>.

Layouts are also grouped by topical sections. Again the Times demonstrates this (top news, analysis, opinion, "the great read", "the athletic", "culture and lifestyle", etc. The Guardian similarly, with several news blocks for top news, "headlines", "in focus", "spotlight", "opinion", "sports", "climate crisis", etc.

HN's news breakdown differs, though looking at the submitted sites, title, and in your case the slug should give some options for largely-automated story placement. I've done my own analysis of HN front-pages, and came up with a list of 47 categories of sites with > 17 front-page appearances (and a great many more without), totaling 16,185 classified sites.

Categories: programming (7719), blog (5506), media (816), science (635), news (344), comm. (227), government (129), software (127), video (78), discussion (73), interest (72), design (60), database (57), cryptocurrency (49), law (41), cybersecurity (25), technology (25), commentary (24), recreation (23), hardware (22), medicine (15), documents (14), military (10), literature (9), economics (8), publications (8), list (7), crowdfunding (6), education (6), webcomic (6), (wiki) (5), books (5), info (5), entertainment (4), environment (4), journalism (4), organisations (3), support (3), information (2), translation (2), humour (1), images (1), n/a (1), networking (1), podcast (1), society (1), ui/ux (1).

(I can provide the classification file on request, username at protomail.com.)

That provides pretty comprehensive coverage of the actual stories submitted (I'd had the exact factor once, I believe it's in excess of 90%).

Again: parsing of the titles and/or slugs (perhaps with an AI assist?) could give better classification. Sites such as Lobsters (<https://lobste.rs>) include tags and often have similar submission selections to HN, which might also be used to organise placement.

Another characteristic of traditional layouts is that the horizontal line is reset periodically. If you look at the NYTimes, Guardian, or other traditional news sites you'll find frequent use of horizontal breaks. I don't know if this is a peculiarity of mine or not, but I find that card-style summaries which are not randomly aligned vertically on a page are much easier to read.

Thanks a lot for the detailed feedback!

Slugs are chosen using AI, but it doesn't work too well, as you mentioned. I will work on iterating the prompt to try and get better results.

A 'remove' option would be a good addition. It would directly remove the element, and not track it as a 'disliked' story.

Since this whole thing is automated, and there's no curation involved beyond embeddings and HN points, it's hard to display it like a traditional newspaper, with proper vertical alignment (since images and blurbs are manually entered to fill the content there).

I think grouping the newspaper into categories is a great idea. I don't think it works with the daily frontpage, since there isn't enough content (just 30 stories on the front page), but it would work quite well in a monthly version. Your categories are very helpful and I will email you about them.

I'll drop you my classification file shortly. It's mostly-manual classification with some pattern-based assignments. Every site that's appeared on the front page a minimum of 15 times, IIRC, and quite a few below that threshold. Coverage is about 65% of front-page articles 2007--2023.

Slug-selection is a tough challenge. AI may not solve it, but at least it will make it more non-deterministic ;-) Reviewing the site again, you're largely doing quite well.

(I'd really like to see a widely-used abstract or summary semantic markup usage, though microformats seem not to have been widely adopted: <https://microformats.org/>.)

The layout / horizontal breaks observation is mostly something that's nagged me for ages in card-based layouts going back to Google+ in the early 2010s. It took me a while to realise that what nagged me about the layout was having multiple columns of cards with no vertical coherence. It just sort of jumbles in my mind, and I'm not sure if that's a cognitive defect of mine or a more general response. I've come to appreciate print-based layout practices increasingly with time, particularly as I find online / fluid layouts increasingly less satisfactory.

(I've made some past observations about layout and on-screen reading on traditional displays vs. a large-format e-ink display. I increasingly strongly prefer the latter for reading, though even there many issues remain. Search: <https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...> and focusing on scroll-vs-paginated displays: <https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...>, if you feel like hunting through a bunch of my whinges ;-)