My favorites are ones which list the most relevant pages it can find (catching scenarios where URLs have changed but no redirect has been added) but only if the site has a functional search feature.
The problem is too many search libraries on sites are anywhere from useless to completely broken, even for some household name brand sites. My go-to test is while on an existing product page to search for either some part of the product title or a snippet of bullet point feature verbatim, to see if the page I'm currently on gets returned at all. It's stunning how many sites fail this test.
Certainly a nice touch. I'm definitely encountering more broken links lately, and it feels a bit off to always blame the client.
A bit off-topic, but thinking about the costs to update a vector search engine VS retraining an LLM makes me a bit worried about hyperlinks all together.
imo it should be considered to achieve the 404 design with minimal complexity. Keep the document small, to prevent i.e. old touch/favicon requests from wasting lots of bandwidth.
The 404 page of the Financial Times is one of my favourites: https://www.ft.com/404
Thank you, that is why I went to the comments.
I appreciate the ones that acknowledge that its probably them that deleted the page and broke the link, instead of saying "looks like you're lost"
My favorites are ones which list the most relevant pages it can find (catching scenarios where URLs have changed but no redirect has been added) but only if the site has a functional search feature.
The problem is too many search libraries on sites are anywhere from useless to completely broken, even for some household name brand sites. My go-to test is while on an existing product page to search for either some part of the product title or a snippet of bullet point feature verbatim, to see if the page I'm currently on gets returned at all. It's stunning how many sites fail this test.
Certainly a nice touch. I'm definitely encountering more broken links lately, and it feels a bit off to always blame the client.
A bit off-topic, but thinking about the costs to update a vector search engine VS retraining an LLM makes me a bit worried about hyperlinks all together.
This is one that I slapped up, many years ago: https://cmarshall.net/Error_404.html (the site, itself, is pretty much moribund).
I took the code from some other site. Can’t remember where, exactly. Like much of Internet history, it seems to be lost in the mists of time.
Put it to good use - using https://notfound.org for your 404 could help find missing kids.
Peugeot.com's 404 page includes - naturally - a picture of Peugeot 404. Unfortunately, there aren't enough Peugeot models for all HTTP status codes.
Their own page is also amusing: https://www.404s.design/hackernews
imo it should be considered to achieve the 404 design with minimal complexity. Keep the document small, to prevent i.e. old touch/favicon requests from wasting lots of bandwidth.
(Not intended to criticize this post)
Happy 404 Day
Yahoo's 404 at Pacific Bell Park was pretty cool.
https://techcrunch.com/2008/08/25/yahoos-404-at-giants-stadi...
In NZ culture there’s a house #404, halfway between the CBD and the suburb, about half an hour by 70 bus.
Somewhere near: https://maps.app.goo.gl/kquCDUWtEWGbp1qX7?g_st=ic
“404 Great South Road, Greenlane, Auckland 1051“