On Friday 20 May 2011 19:23:08 Matthew Toseland wrote: > Another source says the whitelisting thing is not deployed yet, and it'll be > patchy for a while. We'll see. If it is deployed, then IMHO: > - Darknet is essential. > - This means we need really good darknet support e.g. FOAF connections, > invites, i.e. the planned darknet enhancements. > - It will be largely an isolated darknet. > - The restrictions are likely to remain less serious in some places > (western-owned businesses, places where they operate, sometimes academia). It > may be possible to get a VPN through sometimes. It might be possible to link > the internal darknet to the external darknet by linking a VPN'ed opennet node > to an internal darknet node. > - But generally I'd expect *very few* connections to the outside world. > - Hence one thing we should seriously consider is some easy to use tools for > manually migrating content from one disconnected darknet to another, based on > the binary blob infrastructure. IMHO this should include auto-updates, files, > and whole freesites.
One other issue here: If opennet doesn't work, darknet is essential. China scrambles dyndns. So dynamic IPs could be a problem, especially if uptime is limited (it usually is). We might even need some sort of rendezvous transport to make it work - and that would be vulnerable. Aaaaaargh! > > I had a longer discussion and the main conclusions were: > - Freenet 0.5 was very popular in China a long time ago. This explains the > high ranking we achieve on the recent survey. > - He confirms that it was blocked. > - After it was blocked, there was a long period when 0.7 wasn't available or > wasn't ready. > - Right now, most Chinese don't use Freenet because there isn't much > interesting (Chinese) content on it. > - The reasons are similar to why western users don't use it. However, the > history implies that if we can get some of the external chinese content > authors (e.g. chinese language news sites hosted externally) to post content > onto it (or others to mirror said content, which might need better tools for > mirroring), we might get a lot more users fairly quickly. > - Comments on websites are very important - possibly embedded Freetalk. > - Real-time (minutes not hours) comments would be ideal. Some people really > do sit on web pages and read every post - and some of them are important. > - Microblogging is also very important. IMHO Sone is pretty close to this. A > lot use twitter-like tools. > - A mobile version would be good. This isn't likely any time soon for various > reasons (most obviously bandwidth). > - This also implies we need inserting large freesites to be both easier and > more reliable: Currently there are two problems: 1) jsite doesn't allow > persistent site uploads, 2) performance of big sites is poor, and upload > times long, because saces' multi-container code isn't merged. Also, the time > it takes to insert sites is a problem, because it means no quick updates. And > then you get to the isolated darknet issues. > > IMHO real-time is hard. We're not going to get a system that does real-time > with lots of "pages", "forums" or "channels" any time soon, with the > exception of small systems that don't scale such as Flip. However, real-time > in Sone is quite feasible IMHO, because you are only following a limited > number of friends, and we can use CAPTCHAs and relaying for third-party > replies. Full publish/subscribe might allow real-time in Freetalk, but it's a > long way off. > > Real-time systems can work if there are only a small number of people to > poll. This would imply a visible list of who is "logged on" to a specific > chat, with auto-log-off after a period and CAPTCHAs for announcing when you > log back on. I might file a bug for such a real-time-with-channels system. > Obviously you don't need to log on to lurk. This would be slightly more > scalable than current Flip prototypes. > > So, priorities: > - Make Sone official! > - Make Sone subscribe to friends / people you are following at a high > priority. > - Make Sone use CAPTCHAs, or just a low priority subscription, for everyone > else. > - Fix and enable by default saces' multi-container site insertion code. (The > main difficulty is how to make it work persistently iirc). > - Build some kind of tool for large site uploads. jSite can't be easily > changed to support big (persistent) site uploads, so maybe we need something > else. > - Embedding support in Freetalk, suited to comments on blog posts (i.e. > thread-specific). > - Some support for filtering in the site upload tool so we can replace > whatever the markup is for the comments on the external site with the markup > for an embedded Freetalk. > - Think about even better support for big sites e.g. selective reinserts. > (The stuff about big sites is interesting because It's The Content Stupid ... > a lot of content is big sites...) > - Integration between Freetalk and Sone i.e. you should be able to click on a > poster in Freetalk and then either follow them, send them a private message > or whatever. > - Really good darknet support. > - Easy to use tools for content migration. > - Think about writing a mirroring-from-web extension for the large site > upload tools. > > After we have most of this, we should try to get some content i.e. get some > publicity, try to get some China-specific publicity/content, reach out to > them (the obvious ones have mentioned us in the past). > -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 198 bytes Desc: This is a digitally signed message part. URL: <https://emu.freenetproject.org/pipermail/devl/attachments/20110521/02c3e491/attachment.pgp>
