Similar techniques were already used before Web 2.0, but I agree with you, that such methods may increase in future.
Twitter and similar services offer anonymous registration, RSS feeds, live updates in real-time, communication through http(s) - things which are likely to be misused because they fit the bill perfectly.

Though I think that using base64 was not a good idea because it can easily be decrypted.

This herder's communication is fairly obvious, due to it looking like a hash.  But imagine if he/she were to make the bot commands subversive. For example, he/se twits, "Went to the store" and the bot net nodes know that to be "get an update from serverX."

It wouldn't take a lot to do someting like this, it is essentially a simple word substitution and would make it very difficult to tell that botnet communication was happening. To take it a step further, the herder culd rotate the key words/phrases on a regular basis.

The reason I can see for doing this instead of having it hosted on your ISP account, etc. is that ISP's are getting better about shutting down hosted sites when law enforcement calls.  Also, why post your C&C on an ISP known for hosting this type of stuff, when you can fly under the radar on a "legit" site. This method additionally provides availability of a true world class hosted service. Also, if the herder were to setup separate accounts for sections of the net, he could have those accounts follow his main account and post upate resposes.

That's was I meant. There are endless possibilities to to hide the true intends. Masking them so that they look like legit tweets would be one of them.

When offering such a service it must be kept in mind, that sooner or later misusage will occure and that someone exploits it in a way which was not thought of previously.

I don't see much possibilities an owner of such a service have in order to prevent such things. Analyzing the network traffic may help, but again, there are possibilities to circumvent this.