notes.husk.org. scribblings by Paul Mison.

Flickr Catchup: A Technique

I’ve been on holiday for a fortnight, so I haven’t had a chance to check Flickr. How to catch up?

Well, the search box on the “photos from your contacts” page has a handy filter: “from your contacts”. Put in space, press return, and you get… an error. Then go to advanced search, change “date taken” to “posted”, and put in the date you went on holiday, and you’ll end up at a page like this.

(Note the date in the URL; you can change that if you want to skip the navigation outlined above.)

That’s sorted by “interestingness”, so you’d hope the better / more noticed photos would come to the top. If you’re more egalitarian, you could always go for “recent”, and then start at the end of the list and work forward. 

If I was away from a computer more often, I might use the API to turn this into a service, but I’m not, so I won’t. Also, as I’ve noted before, it’s a shame more sites don’t offer filtered search to your contacts. I kept up with Twitter pretty well, but I have no idea what I’ve missed on delicious, for example. Hurrah, then, for Flickr.

Annotations and Machines Tags

I’m at Warblecamp (unsurprisingly, they also have a Twitter account), where I gave a short talk about Flickr’s machine tags and possible lessons for Twitter’s upcoming annotations feature. You can download the slides (6MB PDF), but they’re very much from the “big word / big picture” school, so feel free not to bother.

The idea was to breeze through Flickr’s implementation of tags, machine tags, machine tag extras, and exploring hierarchies via both URLs and the API, and point out the features I liked and how, perhaps, Twitter might learn from them.

The discussion afterwards was interesting. One point, which was well worth making, was that Twitter’s stream of text is very different from Flickr’s archive of photographs. (One more difference is that tags (and machine tags) are editable later; annotations are set in stone at post create time.) Aral Balkan suggested a registry of Twitter annotation namespaces, along the lines of his Twitter Formats proposal. Personally, I prefer the “pave the cowpaths” approach of discovering what’s actually in use in the wild (and is also why I built the machine tag browser). I didn’t mention this at the time, but there was an attempt at a Flickr machine tags wiki, which failed, perhaps colouring my view.

There was also a question about size limits for annotations (turns out it’s 512 bytes) and a discussion on the more RDF-ish aspects of triple tags (and how you say what a thing is, which also touched on establishing concordances). Generally I don’t get hung up on the semantics of machine tags, but I’m sure there are people who do, and they might be reassured by the points (mentioned in the Twitter preview post) about the use of schemas:

People could add some agreed upon “meta-annotation” that points to something which *describes* the annotation or annotations that person is using. Think something sort of like XML DTD, though not necessarily machine readable.

For a few slides knocked up the evening before, I’m vaguely happy with both the talk but very happy with the response and the way it’s made me think more about the idea.