Comment by roc | Hacker News Reader

Comment by roc | Hacker News Reader

Parent Full thread

So do bookmarks. And this overlap becomes obvious as people inevitably ask for Chrome or OneTab to allow them to cluster tabs by topic/project -- because a linear scan is never going to 'work' for people who leave enough tabs open that it becomes a 'problem'.

The underlying problem here is workflow.

People are leaving tabs open as a reminder of things to which they intend to return. (Regardless of whether they will or not; that's another discussion.) And they're not bookmarking, because bookmarking begs organizational overhead, which leads to its own mess. (neither tags nor folders are great or sufficient)

And who knows if a bookmark will still point to the content you intended, when you finally get back to it?

So the problem is ultimately that bookmarking is broken, both for quick reference and longer-term storage. So why not fix that?

Why not a system where bookmarking a site saves a copy to a (cloud-stored) cache. [1] And then searches can be done on the content in that cache. And hits can be served both from the cache, and a simultaneously downloaded 'live' result, available with a toggle. [2]

So that one can bookmark a brag-post about a neat jquery-enabled dropdown list and not have to worry about categorizing it, nor whether or not it will be there in a year, and be confident that they can refer to it again with a simple search of any of the key words that occur to them. [3]

[1] Because one can never know what will happen to content online (changes, broken links, takedowns, etc) and doing a federated search across thousands of bookmarked sites looking for 'jquery dropdown' is going to be a nightmare.

[2] Room here for a great feature of non-trivial difficulty: change-detection and display of diffs (if any) rendered in-line.

[3] Or even searching by meta-data such as date-of-bookmark, location or source-device. "That thing I was reading on my phone last spring, when I was stuck in Chicago on business..." can be surprisingly useful when searching.

Thinking on this more, the cloud-stored cache would wind up eventually being a subset of google's whole-internet-database. From which you could run a pretty awesome search service.

Because it wouldn't consist of randomly spidered or submitted noise. It would only be things explicitly identified as useful or interesting.

And you could rank by how many people bookmarked it, when the bookmarks happened, who they are, etc -- alongside already-useful PageRank attributes.

It'd be like +1 data, but the +1 would have meaning behind it.

Until the SEO optimizers and bot nets themselves start bookmarking their own spammy noise in this cache too. Then we're back to the usual arms race of trying to identify quality content and providers pointing to other quality content and providers while all the junk vendors are trying to masquerade as quality.

Is there a lot of SEO that relies on 'like' or '+1' spam? Because that's the more apt comparison. Particularly when you consider the implementation of even a trivial 'karma' system.

The very creation of a network of mannequins would seem to run counter to the big goal of most SEO (traffic, now) as the mannequins would require months-to-years of seemingly-valid traffic of build-up to become relevant and then a very slow-and-cautious release of bookmarks to the desired spam to even attempt to avoid detection and thus nullification of all the preceding effort.

And the user-expectation of such a search would be heavily weighted to prior personal activity [1]. So even if mannequins were successfully pumping bookmarks to promote some site that just scraped stack-exchange, if you and I were bookmarking stack-exchange, we should always get the real links bubbling to the top.

[1] Yes, this would raise a lot of the same "echo chamber" concerns that people have with Google search, but if it matches the user expectation, I don't see a problem with it.

In passing, one reason bookmarking is broken is that "infinite pages" usually don't return you to where you were, as they should, but to the top of the page.

"Infinite pages" are also breaking search. Google finds a hit but you die of boredom before you can actually scroll down that far....

This.

1. I bookmark lots of sites but rarely go back to them (mainly because it's hard to find what I'm looking for)

2. Sometimes I spend ages looking through my history to find something I remember seeing but didn't bookmark

3. I leave dozens of tabs open because I think I'm going to refer back to them but 50% of the time never do. The order and which window they were grouped with is info I need for context.

So I want -

1. a full-text search of history or bookmarks (I haven't decided which)

2. someway to really quickly convert my open tabs to history/bookmarks

3. A psychic computer that knows when I meant to bookmark something ;-)

I know some extensions can help with this but I've never found the ideal one yet. Suggestions appreciated.

Tracking history would bound right into the 'creepy' range of data collection IMO. But if you were fine with that, it's not hard to imagine such a solution tracking how much time you spent on a given page and using that to make an educated guess that you 'meant' to bookmark it. Or at least weight search results accordingly.

Though the logistics of cloud-storing every single page every user visits would quickly become non-trivial.

And your want #2 seems like a simple UI task. IE already has/had a "bookmark all open tabs"-style menu option -- and I thought the others did too, but I'm not seeing it at the moment.

It's only creepy if it's retained or used against my wishes. In a fantasy world of free data storage, the ability to have a complete history of my life and an ability to search it instantly is incredibly useful. And yes - incredibly dangerous but I'd like the option.

I was looking into this problem lately, the two solutions that come close to what you are describing, from what I found are:

https://kippt.com http://pinboard.in

Pinboard already has archiving, Kippt archiving is in beta.

Still, there's room for improvement.

Fairly certain this is called Evernote :)

Reply on news.ycombinator.com