Wednesday, 4 November 2009

Q: Google Wave Federation, so what?

A: Well by my reckoning quite a lot.

I'll start with my definition of Google Wave: a messaging system. Pretty simple huh, mind you people could argue that it does so much more as a collaborative work tool and that is just the start. On the other hand I have heard it described as Instant Messenger done badly. At its core however it is a messaging system.

Q: So what is radical about that?

A: Nothing really except that it is Federated.

That is to say that it allows multiple systems to talk to each other. This is as opposed to Instant Messenger (IM) which works by connecting through a central point. When you use IM you do not connect to your server you connect to the IM service. Pretty much all of the really popular messaging services work like this except for Email which is the ultimate in federated systems and is the killer app so far of the Internet age.

Q: So is Federation better than Central?

A: In oh so many ways......

The great success if email has been that it does one thing well and in a simple way. It has allowed anyone to set up their own email server and an email server program is often a default installation for server software. This has allowed it to be universally accepted and used as everyone (well the software people anyhow) can tailor it to suit their needs relatively easily. Federation is what has allowed this to happen. Furthermore if your email system fails there are many that will keep going, so federation has built in redundancy. Also, as you not asking one system to do the whole dang you can spread the load and handle loads more traffic, scalability.

These benefits can all apply to Wave as they have already to email.

Q: So why not just stick to email then?

A: Well that is the key question.....

Essentially there is no reason to move to a Wave type setup except for the following reasons, which do not apply to wave alone. 

With Wave you chose to be part of a conversation. You are not forced to receive like you are in email. This means less spam.

It allows you to have linked conversations more easily straight out of the box rather than having to open individual emails and remember what was going on in previous ones.

You can have groups on the fly which can as easily die afterwards with no penalties.

The messages are sent to you rather than you ask to see if there is something new, this is a technical thing but very important since it cuts the amount of network traffic.

You can set up security on it pretty easily as instead of trying to decide who in the whole wide world you might not want to receive from you can decide who you want to be able to see it. This is far more focused and effective.

These may not seem far reaching but I think that they are. What is most interesting to me is that they do not apply to Wave alone. They could as easily apply to an open source version of twitter or facebook for example. They key is federation.

Monday, 13 July 2009

You Can't Keep a Good Idea Down

Throughout recent history there have been many instances of people working in isolation on a problem that is of the mode and coming up with very similar conclusions. Charles Darwin is credited with the theory of evolution but he originally published in 1858 jointly with Alfred Russel Wallace after they both realised that they had similar theories. Radio has a number of claimants to the crown. Powered flight, well it wasn't the Wright Brothers I am afraid to say and don't even get me started on computers.

Theses are all high profile examples but I have seen the same things happen on many occasions in modern computing and in particular with use of the Internet. This has happened to me before and in one sense it has just happened again.

I have already described how a system such as Twitter can be opened up by creating a network of similar micro-blogging services that can communicate with each other and have already made my own test version of one of these.

I also realised that to make it more efficient it needed to have a better way of discovery for the items that were being made public and pushed to a holding system, I called these Directories and I realised that they were capable of applying security and access control at various levels. Users can subscribe to a directory and view items that are applicable to them and for which they have the relevant permission to view.

This type of thing needs to be public domain and I believe that the item needed to make this work was a protocol which we can then build the applications on. Two weeks ago a colleague of mine, Hal Newman, was at a conference run by the US Government on the use of Social Media (OGMA) and I tried to impress upon him the fact that such a protocol was a key element.

At the end of last week I made two amazing discoveries
  1. Laconica the open source micro blogging platform has inter connectivity as a goal but as yet the implementation is somewhat rudimentary
  2. Google has a draft inter operability protocol and sample system in place called PubSubHubbub which can do the kind of thing my directories idea would do
To my mind we now need to bring these two things together. Laconica is a great package that needs more exposure, Google has power and the clout to get this type of thing available. So I am going to try and show how this can happen.

And here is the Google presentation of how this can work:-



More on this as it happens.....

Tuesday, 16 June 2009

Talk About A Revolution

June 15th 2009 was a strange day. Hundreds of thousands, possibly over a million, people on the streets in Iran protesting over the election results that the protesters thought had been rigged. Is this the start of a revolution? I am writing this a day later and I do not know.

What I do know is that the way it was perceived here in the west was very different to how it would have been just a year ago and that was at least in part down to twitter. During the day there were accounts that had decent claims to being genuine voices from Tehran keeping the world updated on events and managing to get photo and video files out to support their claims. So persuasive were these voices that they were getting the traditional news media to follow what they were saying almost in priority to their own on the spot journalists and traditional sources.

This was the true revolution of June 15th. As information was coming out of Iran it was being viewed by thousands and then resent (retweeted) so that it would be viewed by even more. All of this in pretty much real time and it had nothing to do with the large news agencies. In fairness to the news agencies they are beginning to catch on however there is resistance to what they see as their position as the true arbiters of quality reporting.

This is a tweet from Jeff Jarvis from the afternoon of 15th June "I emphasized to a reporter today that Twitter is not the news source. It's a source of tips & temperature & sources. Reporting follows." Also from his site (posted June 15th) comes the following on the way that the NY Times operates "Because The Times’ brand hinges on it as a product that has been curated and edited and checked and polished - note editor Bill Keller’s language on The Daily Show about his package - it finds itself in dangerous territory trying to compete in real time with those whose brand expectations are entirely different."

According to the Bio on his site Jeff is associate professor and director of the interactive journalism program at the City University of New York’s new Graduate School of Journalism. Well I have news for you Jeff, if you think that any news organisation can get it perfectly right and polished you are deluded and yesterday provided the perfect example.

This is what the editor of the curated and edited and checked and polished New York Times had to say in the early hours of June 15th

Leader Emerges With Stronger Hand

Published: June 15, 2009
President Mahmoud Ahmadinejad’s victory demonstrated that he is the shrewd front man for an elite more unified than at any time since 1979.

Whilst you would now have to subscribe to read the article I can tell you that it was proclaiming that it was a done deal and the middle classes were in Iran were resigned to their fate. Just how wrong can you get it.

So here is the news for you Jeff and for all of those journalists who wish to believe that they are a special breed gifted with superhuman insight and the ability to distill a story for general public consumption. The twitter community and its successors will beat you to the story every time. Furthermore, they will be the people who are the experts in the field and, shock horror, they may even be able to string a sentence or two together. Then the story will be out there and if it gains a following it will spread like wild fire.

This does not mean that journalists are an endangered species, just that they are going to be changing the way they operate in the future. As with any source of information there will be the good, the bad and the disingenuous - these will need to be checked and validated.

Following on from that, there will be the need to draw in comment from other domain experts who are not necessarily directly involved in the main proceedings, for example David Miliband, the UK Foreign Minister, was interviewed on BBC Radio 4 in the morning of 16th June for the UK government's view on the events in Iran. This could not have been done via social networking systems on the Internet.

A news organisation and the journalists working for them can act as a ringmaster in an ever changing circus of events. Constantly watching the crowd to see the news as it unfolds and vetting the shouts from the audience to allow those who have something valuable to say to step into the ring whilst at the same time getting involvement from those who can be invited directly to the ring from outside. They can then step in and lead that conversation rather in the manner of anenormous audience participation show. That, I believe is the future of journalism.

Saturday, 13 June 2009

Saving Private Twiters - a Blue Peter Approach

Twitter clients are pretty simple beasts and do not handle a basic need - to save your timeline of status updates. Most will get up to the latest 200 since this is the default maximum that an API call will allow. The API does allow for paging - i.e. getting them in groups of 200, back to a max of 3,200 but I don't think that anyone bothers with this.

To a certain extent this is not a problem especially in the way that twitter is currentlybeing used however it does not take a massive leap of imagination to relise that saving the dang things might just be useful. As part of a much larger project I am working on I have the ability to save tweets and realised the other day that others may well want to be able to do the same thing.

So here is a Blue Peter style approach - take a server with PHP and MySQL and stick one together. The code is a stripped down version of its bigger brother.

It goes something like this
  1. Create a connection object for Twitter
  2. Connect and get the last 200 updates in your friends timeline
  3. Load the returned element into an XML object
  4. Iterate through the XML elements and insert them to a database
Then you create a cron job on your server to connect to the server every few minutes - I do it every 2 minutes and then you have saved your tweets.

The first thing you will need to do is to set up a database table. I use one called UserSavedTweets with the following structure:-

CREATE TABLE UserSavedTweets (
UserID int(11) NOT NULL,

ServerID int(11) NOT NULL,

UserHandle varchar(250) NOT NULL,

TweetID bigint(11) NOT NULL,

TweetText varchar(250) NOT NULL,

TweetName varchar(250) NOT NULL,

TweetScreenName varchar(250) NOT NULL,

TweetPrfImgUrl varchar(250) NOT NULL,

TweetCreated datetime NOT NULL,

TweetURL varchar(250) NOT NULL,

PRIMARY KEY (UserID,ServerID,UserHandle,TweetID)
)

A couple of things to note here. The first is that this table is designed to handle multiple users on multiple twitter type servers, such as Laconi.ca based servers so the the primary key is a compound of the the various ID's I use in other tables together with the TweetID.

As for the TweetID itself you will notice that this is a bigint which more than happily takes care of the Twitpocalypse potential issues - at least for the next couple of months anyhow.

Now we need to have a method to go get the tweets. I found and used with some alteration a twitter connection object from David Grudl called twitter.class.php

Here is the the current code for that which you can save in twitter.class.php


/**
* Twitter for PHP - library for sending messages to Twitter and receiving status updates.
*
* @author David Grudl
* @copyright Copyright (c) 2008 David Grudl
* @license New BSD License
* @link http://phpfashion.com/
* @version 1.0
*/
class Twitter
{
/** @var int */
public static $cacheExpire = 1800; // 30 min

/** @var string */
public static $cacheDir;

/** @var user name */
private $user;

/** @var password */
private $pass;

/** tweet retrieval param */
private $retrieve_count;

/** the server base address for the api **/
private $serverapi;



/**
* Creates object using your credentials.
* @param string user name
* @param string password
* @throws Exception
*/
public function __construct($user, $pass)
{
if (!extension_loaded('curl')) {
throw new Exception('PHP extension CURL is not loaded.');
}

$this->user = $user;
$this->pass = $pass;
$this->retrieve_count = 20;
$this->serverapi = "http://twitter.com";
}

public function set_retrieve_count( $count ){
$this->retrieve_count = $count;
}

public function set_serverapi( $serverapi ){
$this->serverapi = $serverapi;
}



/**
* Tests if user credentials are valid.
* @return boolean
* @throws Exception
*/
public function authenticate()
{
$xml = $this->httpRequest('http://twitter.com/account/verify_credentials.xml');
return (bool) $xml;
}



/**
* Sends message to the Twitter.
* @param string message encoded in UTF-8
* @return mixed ID on success or FALSE on failure
*/
public function send($message)
{
$xml = $this->httpRequest(
'https://twitter.com/statuses/update.xml',
array('status' => $message)
);
return $xml && $xml->id ? (string) $xml->id : FALSE;
}



/**
* Returns the 20 most recent statuses posted from you and your friends (optionally).
* Ammended to allow more status returns - max is 200
* @param bool with friends?
* @return SimpleXMLElement
* @throws Exception
*/
public function load($withFriends)
{
$line = $withFriends ? 'friends_timeline' : 'user_timeline';

$xml = $this->httpRequest("$this->serverapi/statuses/$line/$this->user.xml?count=$this->retrieve_count", FALSE);

if (!$xml || !$xml->status) {
throw new Exception('Cannot load channel.');
}
return $xml;
}



/**
* Process HTTP request.
* @param string URL
* @param array of post data (or FALSE = cached get)
* @return SimpleXMLElement|FALSE
*/
private function httpRequest($url, $post = NULL)
{
if ($post === FALSE && self::$cacheDir) {
$cacheFile = self::$cacheDir . '/twitter.' . md5($url) . '.xml';
if (@filemtime($cacheFile) + self::$cacheExpire > time()) {
return new SimpleXMLElement(@file_get_contents($cacheFile));
}
}

$curl = curl_init();
curl_setopt($curl, CURLOPT_URL, $url);
curl_setopt($curl, CURLOPT_USERPWD, "$this->user:$this->pass");
curl_setopt($curl, CURLOPT_HEADER, FALSE);
curl_setopt($curl, CURLOPT_TIMEOUT, 20);
curl_setopt($curl, CURLOPT_SSL_VERIFYPEER, 0);
curl_setopt($curl, CURLOPT_HTTPHEADER, array('Expect:'));

if ($post) {
curl_setopt($curl, CURLOPT_POST, TRUE);
curl_setopt($curl, CURLOPT_POSTFIELDS, $post);
}
curl_setopt($curl, CURLOPT_RETURNTRANSFER, TRUE); // no echo, just return result
$result = curl_exec($curl);
$ok = curl_errno($curl) === 0 && curl_getinfo($curl, CURLINFO_HTTP_CODE) === 200; // code 200 is required

if (!$ok) {
if (isset($cacheFile)) {
$result = @file_get_contents($cacheFile);
if (is_string($result)) {
return new SimpleXMLElement($result);
}
}
return FALSE;
}

if (isset($cacheFile)) {
file_put_contents($cacheFile, $result);
}

return new SimpleXMLElement($result);
}

}
?>

So now we need the script to use this, get the latest posts and sae them to your table.

require_once 'twitter.class.php';

// create a twitter account object change the first param to you username and the second to your password
$twitter = new Twitter('twitteraccountname', 'xxxxxxxxxx');
$twitter->set_retrieve_count(200);
$twitter->set_serverapi('http://twitter.com');

$withFriends = TRUE;
$channel = $twitter->load($withFriends);

include("dblib.php");

$link = openDB();

// hard coded elements - normally picked up from current user and server usage
$UserID = 1;
$ServerID = 1;
$UserHandle = 'twitteraccountname';

// IMPORTANT - need to get some validation to prevent injection attacks here



foreach ($channel->status as $status){
$TweetID = $status->id;
$TweetText = $status->text;
$TweetText = stripslashes($TweetText);
$TweetText = mysql_real_escape_string($TweetText);
$TweetName = $status->user->name;
$TweetScreenName = $status->user->screen_name;
$TweetPrflImgUrl = $status->user->profile_image_url;
$TweetCreated = date("Y-m-d H:i:s", strtotime($status->created_at));
$TweetURL = $status->user->profile_image_url;

$query = "Insert into UserSavedTweets values($UserID,$ServerID,'$UserHandle',$TweetID,'$TweetText','$TweetName','$TweetScreenName','$TweetPrflImgUrl','$TweetCreated','$TweetURL')";

$result = mysql_query( $query );

if ($result == TRUE){
echo "Inserted ".$TweetText."

";
} else {
echo "No Insertion $query

";
}
}

?>

And thats it for the script. Then on the server get the cron jobs up and get it running.

Obviously you would need to be able to have a way to access them but that is pretty straightforward standard php MySql stuff.

Thursday, 28 May 2009

Tweets For You and Me

Twitter is the current big thing and there are some really good things about it and there are some really bad things about it.

What is great is that it allows anyone to share ideas and information with the world very simply. There is no long and complicated setup and when you have created your account you are ready to find a few friends and off you go. It does take a while to get used to what is happening but not long. The 140 character limitation is actually a benefit as it forces brevity and allows the follower to skip through things pretty fast.

What is not so great is that it is a closed system. Twitter own it and all its content. This is bad for various reasons. First of all we have given ourselves away to Twitter. Secondly we are at the mercy of what they want to do with the system in the future and finally should the system fail temporarily or even permanently then we will be without it.

This is known and there are a lot of pundits such as Dave Winner out there saying that this is not a good state of affairs.

So what can be done. Yahoo have launched a Twitter type system that they are trialling in Brazil but I fail to see how this helps. It too is a closed system and quite frankly does nothing to help. Any system that is closed will create islands with difficult communications. There are already companies that will search across the various Social Networking sites and presumably they will just add these as they come along but that is hardly perfect and not necessarily cheap either.

So how about this as an alternative. We all set up our own tweet servers. Sounds mad huh? OR is it and this is why I think it is a good idea.

It really revolves around viewing a tweet as a message rather than as a posting to a site with servers having a common API. I see 5 main elements to this:
  1. Set up an account
  2. Follow accounts and allow accounts to follow you
  3. Send a tweet
  4. Receive a tweet
  5. Search tweets
Setting up an account. I set up my account on TEMS.CA (great guys with great ideas - I confess I am one of them so maybe I am biased). My tweeting account is then @andrew.tems.ca which preserves the twitter convention and also borrows from the email convention.

Following accounts is similarly easy in that I ask my server to follow @you.yourserver.com. The server registers that following and sends a message to yourserver.com to ask if I can follow you. The answer is going to be yes, no (blocking), maybe (protected) or we don't know that person. Unfollowing and blocking can follow a similar pattern.

Send a tweet. Log the tweet on my home server just like Twitter does now AND search for all servers that have followers of me. It doesn't matter if there is one follower or one million followers from a particular server as all my server has to do is send a message containing the tweet to each of the found servers. For the receiving server it then logs that message as available for its users to collect.

Recieve a tweet. So we now have it that all the tweets from wherever that you follow are on your server. It is now up to your server to decide how it is going to deliver these to you. The current Twitter API is a very inefficient model and this would allow a server to implement more efficient methods and possibly add additional features as it saw fit. For example additional notes, multimedia etc.. This is very powerful as tweet servers would then be competing on the quality and features of service.

Search tweets. This would appear to be the one awkward element however as I am suggesting a common API all that needs to be known are the tweetservers to be searched and then the results aggregated. Furthermore it means that you could give instant priority to searches made on certain servers (your own for a start) and begin to apply more logic to it.

This shows how simple a tweetoshpere for the masses really is and accords with these 10 rules for social networking which I heartly endorse. Also when you consider that a server could also handle private tweet networks pretty straightforwardly this makes it a very attractive proposition for business and organisations.

This is not just theoretical postualting - I am actively building a demonstration system which may even be ready when you read this. Just mail me details if I have not made an additional post.