It’s interesting you believe power users would pay for features which provide a better way to stream and filter Twitter feeds. The biggest complaint with Twitter seems to be the weak signal to noise and the filters and heatmap would certainly guide someone in exploring the conversations on Twitter. If Twitter could grow to become a platform for grouping and finding patterns in disjoint tweets around the same theme or anchor, it could be a good place for data scientists to find an underlying collective consciousness that can be extracted. For an average user though, the most compelling use case probably remains to be what Twitter was originally created for — a platform for asynchronous conversation in which your voice can go viral. If Twitter could somehow improve the chance for those really good (e.g., relevant) tweets to go viral, that would add value. To do that, Twitter has to find ways to improve the signal to noise and refine the definition of relevance from categorically relevant to individually relevant. Chronological sorting and heatmap are not sufficient.