3 ms·

I wonder if you could flesh out your thoughts on Gnip and other licensed API syndicators such as datasift/mediasift.

nikcub 15y ago
Gnip has shifted a bit. When it started out the purpose was to solve 'the friendfeed problem'[1] - which was that for each service friendfeed integrated with, they had to write an entire module to implement a client of that services API The thinking was that Gnip would implement the client once, and then expose a standard interface so that a dev could integrate with all of these services using just a single API The second part was that instead of polling, Gnip would push update data to you using XMPP or callbacks. I think they found that there isn't that large a market for apps, at that time or this time anyway, that needed to integrate with a dozen or more services to make paying Gnip worthwhile The second part solved itself through either pubsubhubbub or API's implementing their own callback. Gnip downsized for a while, but saw a resurgence with the Twitter deal. It makes sense for Twitter to offload serving the API, implementing track, etc. I think datasift is a bit different. Actually a bit more advanced since they support many more ways to slice and dice the data, analytics, and more. That space is more like hootsuite and the similar services - you want to mine social network data and extract key analytics. Gnip is more about making API consumption for apps easier. Gnip has 50+ services, Datasift about 6 or 7 - so you can see where they are beginning to diverge. When Gnip first launched, I spent a day with the team as part of covering the launch for techcrunch. My own conclusion at the time was that Gnip was a stop-gap measure between then and whenever the protocols such as PuSH would be stable enough to see wide adoption. There was also the element of outsourcing your API which didn't seem right - Mashery does something similar with outsourcing dev community, API tokens etc. But Gnip have proven otherwise with the twitter deal. (edit: for reasons why twitter would use them, see [4] What could be interesting is Gnip has morphed into part of the old Gnip, with aggregation and normalization, plus something akin to Cloudflare but for API's - while datasift has become a datamining and analytics platform more for internal business use. The space that Datasift is in will be huge. With Gnip I am still a bit indifferent only because I am implementing an app at the moment that talks to a dozen or so different API's and I couldn't see how I would use it or why I would pay for it, but the twitter firehose isn't one of hte API's (but twitter auth is - and Gnip, from what I know, doesn't touch auth at all) As an aside: and interesting point in history is that when Gnip launched twitter totally rejected them, only later to give them access and then take it a step further with a full commercial partnership[2][3]. that was back when the twitter firehose was 5-6 tweets a second, it hit 5k a second yesterday [1] http://www.techcrunchit.com/2008/06/26/the-new-datastream-aggregators-friendfeed-and-standards/ http://www.techcrunchit.com/2008/06/26/the-new-datastream-ag... [2] http://www.techcrunchit.com/2008/07/18/battle-over-twitter-opens-up-to-gnip/ http://www.techcrunchit.com/2008/07/18/battle-over-twitter-o... [3] http://techcrunch.com/2008/07/18/twitter-plays-nice-xmpp-firehose-data-feed-to-gnip/ http://techcrunch.com/2008/07/18/twitter-plays-nice-xmpp-fir... [4] http://www.readwriteweb.com/hack/2010/11/why-is-twitter-partnering-with-gnip.php http://www.readwriteweb.com/hack/2010/11/why-is-twitter-part...

Gnip has shifted a bit. When it started out the purpose was to solve 'the friendfeed problem'[1] - which was that for each service friendfeed integrated with, they had to write an entire module to implement a client of that services API The thinking was that Gnip would implement the client once, and then expose a standard interface so that a dev could integrate with all of these services using just a single API The second part was that instead of polling, Gnip would push update data to you using XMPP or callbacks. I think they found that there isn't that large a market for apps, at that time or this time anyway, that needed to integrate with a dozen or more services to make paying Gnip worthwhile The second part solved itself through either pubsubhubbub or API's implementing their own callback. Gnip downsized for a while, but saw a resurgence with the Twitter deal. It makes sense for Twitter to offload serving the API, implementing track, etc. I think datasift is a bit different. Actually a bit more advanced since they support many more ways to slice and dice the data, analytics, and more. That space is more like hootsuite and the similar services - you want to mine social network data and extract key analytics. Gnip is more about making API consumption for apps easier. Gnip has 50+ services, Datasift about 6 or 7 - so you can see where they are beginning to diverge. When Gnip first launched, I spent a day with the team as part of covering the launch for techcrunch. My own conclusion at the time was that Gnip was a stop-gap measure between then and whenever the protocols such as PuSH would be stable enough to see wide adoption. There was also the element of outsourcing your API which didn't seem right - Mashery does something similar with outsourcing dev community, API tokens etc. But Gnip have proven otherwise with the twitter deal. (edit: for reasons why twitter would use them, see [4] What could be interesting is Gnip has morphed into part of the old Gnip, with aggregation and normalization, plus something akin to Cloudflare but for API's - while datasift has become a datamining and analytics platform more for internal business use. The space that Datasift is in will be huge. With Gnip I am still a bit indifferent only because I am implementing an app at the moment that talks to a dozen or so different API's and I couldn't see how I would use it or why I would pay for it, but the twitter firehose isn't one of hte API's (but twitter auth is - and Gnip, from what I know, doesn't touch auth at all) As an aside: and interesting point in history is that when Gnip launched twitter totally rejected them, only later to give them access and then take it a step further with a full commercial partnership[2][3]. that was back when the twitter firehose was 5-6 tweets a second, it hit 5k a second yesterday [1] http://www.techcrunchit.com/2008/06/26/the-new-datastream-aggregators-friendfeed-and-standards/ http://www.techcrunchit.com/2008/06/26/the-new-datastream-ag... [2] http://www.techcrunchit.com/2008/07/18/battle-over-twitter-opens-up-to-gnip/ http://www.techcrunchit.com/2008/07/18/battle-over-twitter-o... [3] http://techcrunch.com/2008/07/18/twitter-plays-nice-xmpp-firehose-data-feed-to-gnip/ http://techcrunch.com/2008/07/18/twitter-plays-nice-xmpp-fir... [4] http://www.readwriteweb.com/hack/2010/11/why-is-twitter-partnering-with-gnip.php http://www.readwriteweb.com/hack/2010/11/why-is-twitter-part...