Known Implementations: None
Putting Together Salons (Automatically)
When I refer to a salon, I mean the concept of the salons from the 1800s. Salons were held partly to amuse one another and partly to refine the taste and increase their knowledge of the participants through conversation. (Thanks to Steve Sanderson for suggestion of that term to describe what I am talking about)
The concept here is that you would like to get groups of people together to talk about a particular topic. So you put in a topic and a seed group of people. You also put in the necessary parameters that would be needed for a meeting to occur…# of people, time, place, location. So maybe you have a pool of 8 people who like to talk about startups. You could set up a meeting of 3 for Thunderbird Coffee, Thursday, September 23, at 9am. When the scheduler wakes up, it picks three random people from the seed list for the group and it sends them an email with a 24 hour expiration date. The email has a yes/no/don’t bother me link. If you get 3 Yeses, you send an invite to the meeting and it’s established. If you get 2, then you send a message to one more person until you get another Yes. Then, send the invite. If you cannot achieve a meeting of 3, you send an email to the Yeses and you say I’m sorry, we could not make this meeting happen.
At the conclusion of a meeting that was supposed to have happened, you send a message to each participant.
You ask them to fill out a short survey:
- Did you attend the meeting?
- How would you rate the meeting (1-5)?
- If you could invite one more person to the meeting, who would it be?
- If you did not really click with someone at the meeting, who was it?
The system would use the intelligence of these surveys to make future meetings better by pairing people who seemed to work well together and not pairing those who don’t, etc.
The system would allow a person like me, who often wants to get people together for various reasons, to build small topic-focused interest groups, but also assist me in improving the quality of the groups over time. The randomized nature of it is kinda fun. It’s like a jury summons but without sitting in court. The automation allows for the potential to scale the regular meeting of small focused groups. The seeding is important, but hopefully the continuous feedback from the surveys would help guide group progress going forward.
Notes about this idea’s history:
Many people listen to music these days with the help of the web. Services like Pandora and Last.fm to name two. So our preferences are being recorded already. One problem though is that we may not use the same service so it’s not necessarily easy for me to see what you’re listening to if I use Pandora and you use Last.fm. So we build an app with the APIs of the services to aggregate those. Secondly, there’s a social piece where once you’ve entered your id for the various services, you post to Twitter to invite your friends to this site to hook up their preferences so you can share tastes. Once they click the link and add a service, by virtue of the fact that you are Twitter or Facebook friends, their musical tastes automatically start influencing what you see in your home screen. You should also be able to disable the influence of a particular account or maybe even say “i don’t like this song” and eventually you build up new intelligence about their musical tastes that is original data. Every song should ideally have a playable sample and be linked to one or more sites with an affiliate link. I think it should be like a nickel a song if someone buys, if I remember right. So hopefully, if enough people start using it, you would eventually have more than two nickels to rub together.
If you’re interested in hearing more about this idea, send me an email at ideas –at– damonc dot -com.
Have you ever paged through Twitter friend pages? They use cursors in the URLs for better scalability.
It sucks for two big reasons:
- they only show 20 at a time (with no other options)
- If you go to http://twitter.com/followers and click ‘Next’…then hit ‘Back’. FAIL! The page is blank.
It might be hella useful to have a Gigantic friend (or follower) page which would display up to 5,000 friends at once. Then, implement a bunch of features like Follow, Block, Report for Spam, which lists are they on, add to list, pic, bio & location, etc.
If you need to link in a person’s current avatar in your web app, @raffi added a neat little trick to pull in a couple different sizes. Very handy!
It does a 302 to the current avatar image file on S3.