> Maybe make it heavier cached ("slower" time before new tweets are available, less infrastructure costs) but wouldn't disable it fully.
This. I had a brief flirtation with working with people using the Twitter API for research purposes. A very large majority of them didn't care about having the latest tweets coming off the firehose, but instead getting access to all the old data. At least at the time, the latter was much more difficult to do.
Things like using Twitter for real time sentiment analysis does require realtime data, but those applications are more likely to be for-profit, and thus could afford to pay.
Right, the 30 day search API is still 100 request/mo for the sandbox and 500/mo for the bottom paid tier (which I think is $150/mo). Would love to hear more from you about this experience, whether in blog form or commentary. I use the filtered stream API a lot and the search API a bit.
Oh I’m sure. My point was just that the use cases least likely to be able to afford paying are clustered on historical data. And those clustered on real-time data tend to involve $$ anyways
This. I had a brief flirtation with working with people using the Twitter API for research purposes. A very large majority of them didn't care about having the latest tweets coming off the firehose, but instead getting access to all the old data. At least at the time, the latter was much more difficult to do.
Things like using Twitter for real time sentiment analysis does require realtime data, but those applications are more likely to be for-profit, and thus could afford to pay.