search

hypothesis / h

Annotate with anyone, anywhere.
https://hypothes.is/
BSD 2-Clause "Simplified" License
2.96k stars 427 forks source link

Generate a short url to the overall page. #234

Closed dwhly closed 11 years ago

dwhly commented 11 years ago

Regardless of the obvious issues w/ short URLs, the fact remains that there is a need for them. I use them many times a day.

We should provide a fast and easy short URL service both for overall pages as well as for annotations, we can avoid an extra hop (and an extra archival weakness) for our users, and give them one more reason to use the application.

Possibly we could give people a way to provide an optional comment attached to this short URL link?

Hmm... then it actually becomes an annotation...

tilgovi commented 11 years ago

I don't see the benefit. Given that our annotations' bodies support markdown a link within a body can be given arbitrary link text. Anyone sharing a link to a page on social networks is subject to those networks' rules about links. I would prefer not to see t.co -> hpt.is -> when someone puts something on Twitter. Users may have their own preferences for link shorteners, anyway. Unless we plan to build out stats pages and be some kind of social analytics thing I don't see any reason to have our own short links.

tilgovi commented 11 years ago

Basically, I see short links as serving two functions usually: 1) Shorter URL 2) Click tracking

I don't see a serious need for either for us.

jtremback commented 11 years ago

I agree

On Sun, Jan 6, 2013 at 8:52 PM, Randall Leeds notifications@github.comwrote:

Basically, I see short links as serving two functions usually: 1) Shorter URL 2) Click tracking

I don't see a serious need for either for us.

— Reply to this email directly or view it on GitHubhttps://github.com/hypothesis/h/issues/234#issuecomment-11940881.

dwhly commented 11 years ago

Four kinds of annotations:

I see the last kind of link as essentially a prediction that there is utility in being able to point to a page in a way that's attributable to me via my hypothes.is presence, even if I don't contribute anything (body) directly. In other words, there's value in knowing that someone with a high reputation created a link to something that many people follow.

I have a hunch it's useful-- but I agree that it's probably only so after we're much more mature, and the value of knowing the reputation of the link-creator is helpful.

Back to the argument for a short URL service for the overall page:

You seem to agree on the notion of a short URL for the first one. I assume you wouldn't have a problem w/ short URLs for the 2nd or 3rd either. Your question then is around the value of a short URL for the last element, but your overriding argument is probably that you question the value of us exposing a separate URL for a page that's a substitute for the actual URL.

You rightly point out this ugly chain.... t.co -> random link shortener -> target page

We all agree, It's terrible, and slow. But I seem to do this almost everyday, why? Because twitter has not yet opened their t.co URL wrapper API up to external apps (I just checked again on all the twitter clients I use)

Instead I use the Shareholic chrome extension to generate and copy into my clipboard a goo.gl URL for the page most of the time, because it's easy. I guess I figured it would be handy (and easy) to provide the same service ourselves.

Obviously not the highest priority, but perhaps we can tuck the issue up in the attic for now.

tilgovi commented 11 years ago

Yep. I'm with you. I'm for the first three and against the fourth.

My argument against the fourth is that the link itself need not be attributable to an individual if the conveyance is For example, when I share a link via Twitter link need not be "mine" if the tweet is. We can presuppose some indicator of my H. reputation wherever my Twitter name appears to address the use case you present about reputation of link sharers as interesting metadata (e.g., an 'automatic' annotation of content which unambiguously identifies an individual known to Hypothes.is showing that individual's reputation; such unambiguous identifiers might be text marked up with schema.org metadata linking to a profile which is connected with one in H. or FOAF microdata, etc).

Links are just URIs conveyed (when URIs are placed in a document they become links). We should not bifurcate the URI in order to assign an owner when the vehicle of distribution is an appropriate point to attach authorship which does not do damage to the resource identity space.

The only counter-argument I see arises when click-tracking is desirable. I argue that such tools exist and unless we want to expand our scope to include these kinds of social analytics tools we should stay away from creating short URLs. Users interested in tracking clicks through the links they share can use existing tools to do so.

It's not even clear that the first three use cases are properly "short URLs" in the common sense. They could be short (in length) indeed, but I would propose to make them alternates rather than redirections: http://hpt.is/a/XYZ owl:sameAs http://hypothes.is/annotations/XYZ rather than http://hpt.is/a/XYZ (redirects to) http://hypothes.is/annotations/XYZ.