Really a better caption would be “Maybe a bird was eating it so google stepped in and ruthlessly ‘euthanized’ it with a large rock”.

I first wrote this post as a reply to his blog post, but it was a bit longer than I intended and I have written nothing here in the recent past.

I disagree with him for two main reasons.

Firstly, good twitter clients make hash tags into links to a twitter search for the hash tagged word. Tweet deck is a good example (and if you haven’t migrated to it from Twitterrific yet, you should at least try it). This is obviously a convenience factor, and an easy way of directing people to an existing conversation on a topic. Remember that the @ tag started as a user meme which was eventually incorporated into the actual service.

Secondly, and perhaps more importantly, hashtags are usually how people show that the mentioned tag is the main point of their tweet. So, even if the search function would give the same result, a hashtag allows me to quickly see that a word is not just mentioned in passing.

Tangentially, hashtags have now become a meme/convention. To not use them is to allow your message to be lost in the unsearched ether as there are now plenty of tools which search for only #hashtags – especially notably hashtags.org. Whether or not you like that this has happened, it is now the case. It seems silly to ignore it.

Does anyone else have an opinion?

Purpose:

• Allows mass invisible messages by reposting a direct message to it to all those following it. (This is invisible as long as the bot itself is ‘protected’, which it is.)
• Effectively creates a ‘group’ within Twitter.

Usage:

• Follow @spsp
• Direct message your message to spsp with the regular syntax. i.e. “d spsp @zethraeus I am testing your excuse-for-a bot”.
• The message will be reposed as a regular tweet by the bot, without the initial ‘d sps’, but with your twitter alias signed at the end. i.e. the above message would be posted as “@zethraeus I am testing your excuse-for-a bot ~ @alias”.

Limitations:

• The bot can not handle accented characters. It will cut off your message from the point the accented character is introduced. This is the case because of a quirk in php’s libxml2 functionality – and because I am not a particularly good or dedicated PHP programmer. This quirk is not a bug (apparently).
• Should your message be above 140 characters once the ‘d spsp’ is removed and your alias is appended, it will be shortened in Twitter, and any clients like Twitterrific. It will require a click through to see the unabridged version. Whilst this isn’t a huge issue, it’s probably worth keeping the tweets a little bit shorter.
• In theory, if the modified tweet exceeds 160 characters, it may not send. In reality, Twitter wont let you have a name long enough to make this occur in normal usage.
• The tweets are passed on every five minutes. (This is a cron job)
• The maximum processable tweets at each 5 minute slot are 20. I don’t know enough about the workings of http://twitter.com/direct_messages.xml to say whether any tweets exceeding the 20 will be lost into Twitter or just processed 5 minutes later.

The script is comprised of three files.

1. process.php grabs all of @spsp’s direct messages since the last recorded id, records the new last id and reposts the direct messages as tweets.
2. twitter.php, found at http://www.aimclear.com/code/twitter/twitter.txt is a modified version of the twitter api. It is responsible for actually posting the status.
3. lastid.txt is simply used to record the id of last read direct message so that tweets aren’t posted twice.

All of these files are available for download and the bits that are mine, are in the public domain for you to do what you want with.

I just hope that people use it!

A couple of months back, I had an idea which could potentially increase Twitter‘s worth to me.

Up until a month ago, I attended school with, among many others, a small group of rather close-knit twitter users (euphemism aside, I mean geeks). We used and still use Twitter as a passive medium for random comments, status updates and importantly chat. One downside that I have always seen in using Twitter, especially as an open chat client, is the inability to divide your followers into groups and serve them content separately. For Twitter to implement this would be to completely betray the main purpose of their service, but still I have sometimes wanted to send a message to only my school friends. I wanted to a twitter bot (wrong word?) which would make that possible.

The basic idea of the bot would be for everyone in the ‘group’ to follow a protected bot user. Then in order to message only the others in the group, they would send an invisible direct message to the bot, which would then relay it in its entirety to all of its followers. In this way all of the accepted followers of the bot account would get the message, but it would be delivered invisibly to the outside community. The nearest you could come to achieving this goal in twitter otherwise, would be to directly message every ‘group’ member individually.

So, today I took a quick dip into Twitter’s API to see if my vision was workable. Conclusion: using a php server running a cron job every 5 minutes, it would indeed be possible. And even I should be able to do it. The two main downfalls are that messages could only be relayed in 5 minute intervals, due to limits associated with querying the API, and should more than 20 direct messages be received within a 5 minute period, the earliest would be lost in the nethers of Twitter’s dark underbelly. (Twitter will only ever serve you a maximum of 20 direct messages at once.)

Interaction with Twitter’s API can be done through the reading of and posting to special XML and JSON files, and in some cases RSS and ATOM feeds.

So, would anyone actually use such a system if I were to hack it together?

Resources:

• Official Twitter API documentation
• A basic method to preform a status update with PHP
• Parsing XML with PHP (with SAX)