Sunday, April 11, 2010

Because I always need a new project...

The following was sent off to the Twitter folks to get their feedback:

#941064 Concerns about a new application I'm developing

I've prototyped an application which is designed to allow -- essentially -- plays acted out over Twitter. The idea is to essentially have a set of characters twittering from some altered reality. The idea is that I could put on a "War of the Worlds"-type drama (or perhaps a zombie invasion) using a small handful of Twitter accounts.

The application essentially runs a script, in a manner you would expect given the desire to put on "twitter dramas":

userid TIME-SPEC message

The definition of TIME-SPEC is sufficient to allow both exact times, as well as offsets from the previous message, so it should be both accurate enough and not clunky.

What I have now works -- I've done just enough testing to know that -- but I want to make sure that I neither violate any Twitter rules, nor make it easy for spammers to abuse my application.

1. Is the very concept of this something expressly not allowed?

2. On "serial accounts": To perform a twitter drama, multiple accounts are required. One account for each character -- otherwise the effect just doesn't work. However, to follow the drama either each post needs #marked, or the "viewing public" needs to follow every account. I think a larger cast with #ImATwiterDrama breaks the feel when compared to a smaller set of actors that the viewers need to individually follow. I think the strength here lies in limiting the number of characters. Allowing that writers can work within reasonable limits, what would be a reasonable limit to keep the spammers at bay?

3. Since these tweets would be a work of fiction, limiting their ability to post to trending topics makes a lot of sense. I can see the possible need for #ShowName, but other than that they shouldn't be allowed to #mark posts to influence trending topics.

4. The Twitter platform provides a unique method of integrating "viewer feedback" in to the show. While most of the replies will be between characters, real people will be @replied to -- but this is only allowed when the @human is currently following the character. This puts the human in charge, and should prevent reply-spam.

5. Regulating the number of tweets a day: This application makes tweets, but isn't designed to read tweets. (Reading tweets would be done by the author using their normal application.) The number of tweets normally allowed is pretty large. Should I add an application-side limit on the tweet frequency, or is this not needed? If a story has too many tweets it can start becoming clutter, but a slow build-up may lead to a couple days of frenzied activity. I just don't want to be too attractive to the spammers.

6. Absolutely no support for following people. If the character's accounts are to follow anyone it needs to be done by the author using their normal software. The goal is to get people to follow these accounts, not for these accounts to follow people. Also note, since the software should only allow @replies when the targeted human is following the character, allowing the character to follow a human does nothing as far as the software is concerned.

Since this program is essentially an automation tool, things need to be thought-out well before it is released so that it doesn't become attractive to the spammers. I know that. Once I had the proof-of-concept done I became upset as I realized how it could be abused. I still think it is an interesting idea, though, and provided it doesn't implicitly violate the terms of service, I would like any feedback you may have which would allow me to do this while still evading the spammers.

Thank you.

No comments: