<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	>
<channel>
	<title>Comments on: Thoughts on Twitter architecture and pricing</title>
	<atom:link href="http://www.socialstartups.com/2008/05/26/thoughts-on-twitter-architecture-and-pricing/feed/" rel="self" type="application/rss+xml" />
	<link>http://www.socialstartups.com/2008/05/26/thoughts-on-twitter-architecture-and-pricing/</link>
	<description>All that's new in the social computing space.</description>
	<pubDate>Thu, 20 Nov 2008 11:19:40 +0000</pubDate>
	<generator>http://wordpress.org/?v=2.6.3</generator>
		<item>
		<title>By: dlifson</title>
		<link>http://www.socialstartups.com/2008/05/26/thoughts-on-twitter-architecture-and-pricing/#comment-798</link>
		<dc:creator>dlifson</dc:creator>
		<pubDate>Thu, 29 May 2008 03:13:46 +0000</pubDate>
		<guid isPermaLink="false">http://www.socialstartups.com/2008/05/26/thoughts-on-twitter-architecture-and-pricing/#comment-798</guid>
		<description>Andrej,

I've been wanting a corporate Twitter for a long time. Actually, I persuaded Sean Blakey and Alex Cox (SDE friends in Community) to work on exactly that. Sean did his part and built the backend, but Alex and I never got the front end off the ground. 90% of success if execution, I guess.  

So yes, I completely agree.</description>
		<content:encoded><![CDATA[<p>Andrej,</p>
<p>I&#8217;ve been wanting a corporate Twitter for a long time. Actually, I persuaded Sean Blakey and Alex Cox (SDE friends in Community) to work on exactly that. Sean did his part and built the backend, but Alex and I never got the front end off the ground. 90% of success if execution, I guess.  </p>
<p>So yes, I completely agree.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Andrej Gregov</title>
		<link>http://www.socialstartups.com/2008/05/26/thoughts-on-twitter-architecture-and-pricing/#comment-791</link>
		<dc:creator>Andrej Gregov</dc:creator>
		<pubDate>Wed, 28 May 2008 15:31:33 +0000</pubDate>
		<guid isPermaLink="false">http://www.socialstartups.com/2008/05/26/thoughts-on-twitter-architecture-and-pricing/#comment-791</guid>
		<description>I don't think Twitter should charge for their consumer service. They should have some sort of advertising model (maybe something similar to how Twitterific pushes ads into your Twitter stream). They should also launch a corporate product. Allow companies or groups to create private Twitter networks. Something like this: http://en.blog.wordpress.com/2008/01/28/introducing-prologue/</description>
		<content:encoded><![CDATA[<p>I don&#8217;t think Twitter should charge for their consumer service. They should have some sort of advertising model (maybe something similar to how Twitterific pushes ads into your Twitter stream). They should also launch a corporate product. Allow companies or groups to create private Twitter networks. Something like this: <a href="http://en.blog.wordpress.com/2008/01/28/introducing-prologue/" rel="nofollow">http://en.blog.wordpress.com/2008/01/28/introducing-prologue/</a></p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Kyle Mulka</title>
		<link>http://www.socialstartups.com/2008/05/26/thoughts-on-twitter-architecture-and-pricing/#comment-789</link>
		<dc:creator>Kyle Mulka</dc:creator>
		<pubDate>Tue, 27 May 2008 15:29:15 +0000</pubDate>
		<guid isPermaLink="false">http://www.socialstartups.com/2008/05/26/thoughts-on-twitter-architecture-and-pricing/#comment-789</guid>
		<description>I don't claim to know the Twitter architecture, but I can guess where it has gone wrong. 

When a tweet comes in, they give it a unique id, and store the contents and meta-data in a data store. This data store can easily scale horizontally. Your key is the message id.

Then, you copy that message id into each user's mailbox who is following that user. Yes, you have to copy every message id to the mailbox of every user who wants to receive it, but this is a lot better than polling every person you want to follow every time your client wants to update itself.

This way, when a user goes to their page, there isn't any actual work done to figure out which messages to show. They are already in the user's mailbox.</description>
		<content:encoded><![CDATA[<p>I don&#8217;t claim to know the Twitter architecture, but I can guess where it has gone wrong. </p>
<p>When a tweet comes in, they give it a unique id, and store the contents and meta-data in a data store. This data store can easily scale horizontally. Your key is the message id.</p>
<p>Then, you copy that message id into each user&#8217;s mailbox who is following that user. Yes, you have to copy every message id to the mailbox of every user who wants to receive it, but this is a lot better than polling every person you want to follow every time your client wants to update itself.</p>
<p>This way, when a user goes to their page, there isn&#8217;t any actual work done to figure out which messages to show. They are already in the user&#8217;s mailbox.</p>
]]></content:encoded>
	</item>
</channel>
</rss>

<!-- Dynamic Page Served (once) in 0.856 seconds -->
