Well here's my answers: On Mon, Jun 8, 2009 at 11:15 PM, Doug Williams <d...@twitter.com> wrote:
> Jesse, > Please submit an issue if you feel that this would contribute to the > community. There are issues for paging bugs with the social graph methods so > star them appropriately. > Will do - I like to present here first because it opens it up for others to also share their issues and ideas, and helps us get to the bottom of the underlying issue. > I have some questions to the community at large using the social graph > methods so please feel free to chime in: > What is your caching scheme? How dependent is your data on being real time > and why? What type of value are you generating from this data? > I cache all followers, and when looking for new followers compare the followers list with that of my cache to find the differences. I also keep a cache of the individual users. All this plays a huge part in helping others learn what spammers are following them so they can take action appropriately. We also provide several blacklisting options for each user to help prevent spam at the discretion of each user. The goal is to provide as much information as possible about the actions of their followers - the big brands on Twitter are all interested in this. If you contact me privately I would be happy to share with you a good list of those brands using this and interested in its value. Real-time social graph changes are important - by knowing when a user follows another I can know what Tweet could have been the reason of the follow, where they were located at the time, the exact time of the follow, and keep a history of such. If I notice multiple follows by the same user in a single day I can know very quickly whether to flag them so they can decide whether to mark them as spam or not. > What is your use case? There is interest in popular users' social graphs > but from what I've seen they are rather edge case in terms of the value the > contribute back to the community. A valuable use-case outside of a very > specific need would help in prioritizing requests like this. > In my case there are major celebrities, charities, and brands who want their followers to a) be able to contact them, and b) make their followers feel at least some semblance that they care about them. A simple follow, while in reality not much in meaning, means a lot to those being followed. People want to be part of these individuals networks, and being followed by that individual gives them a sense of belonging. In addition, these users want to learn more about the people that follow them. Keeping a list of the new and the old and history and analytics around these users is very powerful. All this is very important to them, to the extent they are willing to pay for it. > > I'm trying to understand where you (the community, not just Jesse) are > generating value from these large follower lists so please feel free to > chime in if you are doing projects on top popular users. > In my case I actually have people willing to pay for this information, so the value to me personally is part of my business model. The users themselves also take value from this for the reasons I stated above. In fact, the more popular the user, the more important this becomes because it becomes much, much harder to manage themselves. Jesse