Comment by jlokier | Hacker News Reader

Comment by jlokier | Hacker News Reader

Parent Full thread

It's completely, totally and utterly different from browser history!

If I visit HN, browse a bunch of comment threads, and find something I want to come back to later, that's 1 tab but 20 history entries.

Same if I searched for something on Google and eventually found something I want to come back to later. Except that's 1 tab and 50 history entries.

To recover that state via history is practically impossible. It's a needle in a haystack. I've occasionally needed to do it, and it takes a long time.

(Compared with a tab, a bookmark doesn't work well either for something to come back to. Anything where there is a context, such as scroll position. Perhaps if bookmarks were better, keeping track of scroll position and allowing text to be highlighted or a note to be attached, and showing a thumbnail (scrolled to the right place), and retaining some temporal relationships between different pages. But I find bookmarks consisting of just a title and URL to be completely useless.)

I would propose that the issue is that browser history navigation is generally poorly designed, at least if you want to do something more sophisticated than hit “back” or search it as a set of URLs. It should be richer and store much more details about each entry.

I'm not sure how to improve history navigation to help with this, as the browser doesn't know which 1 in 20 history entries are worth finding again, and which are junk traversed on the way to something worth finding.

It can't figure it out from content search, because the junk is often the stuff that resulted from useful searches and was assessed by the user as not worth keeping. Keeping tabs is a signal from the user about which entries were useful. A form of curation. Without that signal, what is there?

However, I think bookmarks would be profoundly transformed from nearly useless to better than tabs by improvements to bookmark navigation, UX and a little extra context.

So I have to give credit to you for the idea - history navigation improvements might be transformative in a similar way.

It shouldn't even be a "hard" problem any more. Search and relevance are mature technologies. You do need to provide some features, but there are many: time spent interacting with tabs, was it a "terminus" in a navigation session, active user marking/tagging, signals from other users, other products, etc. It would be an excellent use of ML. The rest is capturing these in a user friendly product, though if it makes Megacorp no more ad money, I doubt anyone will work on it however useful it may be.

Something like this, evolved, perhaps?

https://getmemex.com/

Reply on news.ycombinator.com