Why APIs Are So Cool, by Jeffrey Veen

Why APIs Are So Cool

I've written before about the del.icio.us social bookmark manager, and how to take the data you store there and repurpose it in various ways. This is possible in large part because the creator of the app opened up the entire thing via an API. That means that everything you put in there can come out again, whenever you want, in a standards-base and structured way.

Case in point: Matt Jones pointed me to extisp.icio.us, an example of how an open API is like throwing seeds on fertile ground. Kevan Davis spent an afternoon writing an app that takes a user's del.icio.us categories (or, tags) and creates a visualization based on how many links are stored in each one. Awesome.

The possibilities are endless. Why not show me the overlaps between my tags and those of my friends at Orkut. Ah, but we can't. There is no Orkut API. You can't get that data out of the service once you put it in. Their Terms of Service makes this perfectly clear that you can't use "any robot, spider, site search/retrieval application, or other device to retrieve or index any portion or the orkut.com service."

Apparently locking in customer data through closed APIs and obscure file formats has become an appropriate business model. That's too bad. Just think what we could do just by sharing.

This entry was written by Jeffrey Veen and posted 7 July 2004 at 9:58 AM. It was filed under Technology.

Comments

1. On 8 July 2004 at 1:25 PM dreww wrote:

if orkut's TOS is what it takes to prevent painfully braindead stuff like this, then so be it.

the tags with more stuff in them are bigger, and all are randomly distributed. great. this provides absolutely 0 insight beyond what the normal delicious interface already gives me.

and he's scraping, not using the API, because currently all of the API functionality requires a username and a password and returns data only specific to that account.

2. On 8 July 2004 at 1:29 PM Chris wrote:

Ummm, Kevan isn't using the API. The API requires the username and password for access. This is using a screen scraper, which isn't the nicest way of getting the information.

Kevan should contact Joshua and try to work something out - at least it caches the information for a day, but it's an unexpected load. The about page on delicious is pretty clear about 'not being evil', and Joshua's the kind of guy that would help if approached with a good idea.

Note also that it currently doesn't cluster; it's random placement of words. Using the API, proper clustering could happen.