From: David A. B. <dbu...@jc...> - 2011-08-10 23:44:38
|
Kurtis - On Aug 4, 2011, at 11:15 PM, Kurtis Heimerl wrote: > Hello Papa Legba... ers? Legbites? Legbites is good. > > As you all hopefully know, we at TIER (tier.cs.berkeley.edu) will be > running an experimental service using the Papa Legba network this > upcoming burning man. As we begin to build the system, we had a number > of questions start to pop up, mostly in terms of scalability and > potential user base. I was hoping to get some directions on these. > Firstly, the system as we envision it: > > Basically, users call into a service (at some phone number) and are > connected to other burners. Pretty much a voice chat roulette for > burning man. The secret sauce (and the experiment) will focus on how > much information about the other party we give users, and how much > choice we give them in finding a partner. Those details are somewhat > unimportant though. Great, because every time I tell someone about this system they say "connect me to new people on the playa". Really. > > Firstly, I've never been to BM. Neither has my partner Shaddi. Is this > going to work? Is this going to be useful? What are your thoughts on > the application? Like I said, this is the most commonly requested type of application. Really. > > Secondly, assuming it's the best idea ever, what's a good name for it? > Is there anything similar at burning man (a system for connecting to > new people) that we can name co-opt? We're thinking "phoneroulette" or > " burnerroulette", but I wanted feedback from experts. There is nothing similar. Except maybe the human car wash, were two lines of strangers give each other sponge baths. > > Thirdly, what is the expected base of users? I'm sure you folks have > down and thought about the potential need when allocating towers. What > were you thinking? What's the usual distribution, I assume a lot of > early load as people experiment, with that quickly dropping off. I think a lot of early people do indeed experiment and then drop off. Part of the problem though is that the system gets so congested by Thursday night that it's not very usable. The congestion is not from people actually using the system, but from handsets jumping on and off the network and running constant location updates. By keeping out of the normal cellular bands, this year's system will be more reliable and also require more deliberate effort to access. This will almost certainly change user behavior. > > Lastly, our design currently places users on hold until a sufficient > number of other burners are available to connect to. This means that > we could have 20 or more concurrent users on hold in the system. I'm > certain this load wasn't anticipated by anyone, and is potentially a > problem. At what point do you folks think we might be approaching a > problem? I counted 13 ARFCNs, so that's 90-odd concurrent voice > channels, most in 5 ARFCN bundles. Do we expect much contention for > these channels? How many are we going to allocate as SMS channels? 20 calls on hold is a lot. How long are you planning to hold them there? It's not just a question of capacity, but user patience. There are a lot of (great) distractions out there and you can easily lose someone's attention in just a few seconds. Is there any way to do this with a callback mechanism? > > Hopefully this will jumpstart some discussion. We'll have a prototype > up and running online here shortly as well, I'll send that out when > it's available. > > Thanks! > > ------------------------------------------------------------------------------ > BlackBerry® DevCon Americas, Oct. 18-20, San Francisco, CA > The must-attend event for mobile developers. Connect with experts. > Get tools for creating Super Apps. See the latest technologies. > Sessions, hands-on labs, demos & much more. Register early & save! > http://p.sf.net/sfu/rim-blackberry-1 > _______________________________________________ > Openbts-bm2009 mailing list > Ope...@li... > https://lists.sourceforge.net/lists/listinfo/openbts-bm2009 |