I have been pruning my Twitter follow list over time. It was initially at around 160, then it dropped to about 130 and it was at about 105 till yesterday. Last evening, I gave it one almighty chop and pruned it further to about 66.

The chops were done largely on the basis of the following rules *:

No automated feed publishing on personal accounts

No high-volume automated accounts (Example @techmemeFH)

No compulsive retweeters (Example @timoreilly)

No mutual-hugging A-Listers.

People I don't really know, who say things I am clueless about

People who have not updated in yonks

* There ARE exceptions to the rule.

Quite a few of these rules come from a simple issue I have with Twitter. That there is no way at this moment to vary attention spent on each tweet. The problem ideally should not exist with Twitter. At 140 characters, it should be easier to consume/follow/track. But 140 characters also mean there is little skipping of text, much fewer patterns to be matched, you wind up reading more than less with Twitter.

Let us assume that each tweet costs you one attention unit to read as it scrolls past your screen. Unless you are not online with a client or not on the Twitter website itself, there is no way you will wind up spending less than that one unit per tweet. On the other hand, there is always the potential of expending more than one unit, if the tweet is actually interesting.

The shorter version of the story is that the efficiency of using Twitter comes tumbling down very fast if you wind up following a lot of people or people who consistently tweet a lot. There are services like Filttr that promises to help deal with this, but Filtr's filtering has left me more confused and clicking filtered tweets, which gets me spending more time than earlier with the Tweets

The other part of the problem comes from what I have mentioned before: Twitter's discovery model, either for people or for content is largely broken. You cannot discover other interesting people if you don't follow people who are interesting to you. This is horribly manual, inefficient and after a while comes back to bite you on your ass, because without constant pruning, you will wind up following a lot of people for they may say something of interest to you some day.

You can address the problem to an extent with hashtags, but the volume there is simply overwhelming and creates another problem because of that. And the issue is larger than just Twitter or the client apps.

Google has, as an unintended consequence of its generic search working well, mucked up automated content discovery on the net on an individual level. Generic search works by mapping behaviours and patterns in linking and the clicks that result from it, which is then reduced to identifiable patterns. Suggestions are then based on the pattern matches, than on the item itself. Suggestion for Google is a quantifiable similarity in a fingerprint. This works fine on a 30000 feet level, at 100 feet, it is an unmitigated disaster.

The trouble is also that Google's computing infrastructure is hyper tuned towards supporting such a reductionist system, leaving them little leeway to tackle this problem on an individual level. It is a system that addresses needs well if they are generic enough, but with fringe cases it can and it will leave you feeling that it is just not good enough. What would solve the problem are tiny Google systems working only within the data set which is why universe of data and the way I identify and relate to them.

As you can imagine, such a 'mini' system is largely a fantasy because systems that work well over huge volumes of data tend to break down or become useless in what they output when they are given smaller data sets. Simple fact of life for such systems is that they need to be 'taught' a lot first before they can actually learn anything about you. For the time being, I will have to stick to keeping a smaller and manageable follow list as a result.

The downside is that I will be chopping off the ability for 'network effects' to kick in as a result of my pruning. But I will, to make up for that, make an effort to try and respond to all the replies I get.

I will keep you all posted on how it works out.