About Twopcharts

Home

Deutsch

To be considered for the Twopcharts, a tweeter needs to tweet in the relevant language and needs to be active. It is not necessary to tweet exclusively in the language, or even have the majority of tweets in the language. As long as regular tweets in the language can be observed, a tweeter can be considered. An active tweeter is someone who tweets on a regular basis, and has not stopped tweeting. If the frequency of tweets falls below a certain threshold, or if no new tweets are observed, a tweeter will automatically be taken off the list. The current limits are not too restrictive. It is not necessary to tweet every day, and a break for a vacation, or trip or any other reason, will not lead to immediate removal from the list.

Once these criteria have been met, ranking can take place. The basic idea of twopcharts was to find the 1000 tweeters who would be the most interesting to follow. An obvious measure for this is to track the number of followers a tweeter has achieved. Unfortunately, just checking the amount of followers can give a distorted picture, because of a combination of courtesy of tweeters and availability of mass follow tools. Being a social platform, there is a strong tendency of tweeters to check and follow back tweeters who have started following them. With the availability of tools that allow tweeters to mass follow hundreds of other tweeters with the push of a button, these mass followers are merely trying to attract followers to their account. Often this has little to do with social interaction or tweeting interesting messages. In many cases following these mass followers would just be a waste of time and clutter your timeline. Tweeters who use these practices are often recognizable, because they often have more following than followers, with high numbers for both. Twopcharts is trying to filter out mass follow accounts. It does this by assuming that the amount of tweeters that one follows gives an indication if a tweeter is genuine or a mass follower.

It is assumed that it is almost impossible to actively follow more than 1,500 tweeters. If someone follows more tweeters than this, it becomes more likely that at least some mass following practices are taking place. When a tweeter follows more than 3,000 other tweeters, it is assumed that all following is related to mass following. What this means for the twopcharts is that for all tweeters that follow more than 3,000 other tweeters, the total amount of following is subtracted from their total amount of followers. When someone follows reasonable amounts of other tweeters, only a small correction is made to the amount of followers. Around the maximum assumed reasonable number of following 1,500 other tweeters, a correction is made of 500. The main reason for doing this, is in order to be able to compare different tweeters with the same amount of followers. Simply said, for comparability reasons it was stated that a tweeter with 0 following and 1,000 followers is as interesting to follow as someone with 1,500 following and 1,500 followers. This is obviously a subjective assumption, which may not be entirely correct and the assumption may have to be revisited when more experience is built up with the charts. The correction factors that are applied can be shown in the following chart.

Twopcharts
It can be observed in the chart that certainly up to 1,000 following very little correction is applied, and since all tweeters are subject to the same ranking methodology comparability should work reasonable well.

To further illustrate how this works, we can run some numbers. Someone with 500 following and 500 followers will have an adjusted follower count, which is used for the ranking, of 471. With 1,500 following and 1,500 followers, the adjusted number is 1,000, and with 3,000 following and 3,000 followers it is 0.

Someone with 4,000 following and 5,000 followers will have a ranking comparable with someone with 1,500 following and 1,500 followers, or 0 following and 1,000 followers.

If all this is fair is obviously debatable, but it does achieve a number of things:

Someone who has started using twitter and is starting to build up a friends list is hardly penalized. With 500 friends, the model assumes almost all to be genuine, with the difference only used for comparability reasons.
Tweeters who have a wide network of contacts are not much impacted compared to each other. Someone with 1000 following and followers will have roughly the same rank as someone with 2000 following and followers, and they will appear close to each other in the ranking.
A lot of mass followers do not appear in the list. Every tweeter with more than 3,000 friends and more friends than followers will not make the list, because they will have negative ranking points. In most cases I think this is a good thing.

Like every model there are obviously also flaws, but since the model has to be workable, not all exceptions can be taken into account. It is possible that tweeters are being followed because they are of interest to someone, and that they follow back out of courtesy. These tweeters can build up a large number of followers, but at the same time increase there own following count. When the numbers are small this is not a problem, but when they get large, this tweeter will not appear in the list, even when he/she should because there is apparent interest in his/her tweets. I estimate that at this moment I am only missing a handful of people because of this. My advice would be to just bring the following list down to tweeters that are actually of interest. This does not mean at all that less communication is possible. There are many tweeters with small following lists and large follower list who constantly engage in dialog. Through replies and mentions these tweeters interact constantly with other tweeters, without the need to follow them.

In general I don’t think that anyone should focus too much on exact position in the twopcharts. There are many active and interesting tweeters, all worth taking a look at, either to follow their tweets, or to start interacting with them. This is the main reason I chose to go for a list of 1,000 tweeters. Not to determine exactly their place in the twop-1000, but to make sure I found a lot of interesting tweeters.

The Tweet-1000 is derived from all tweeters that are considered for the Twop-1000. It is not a separate search through the entire universe of tweeters, nor is it limited by the 1000 highest ranking tweeters. In order to compile the lists, an increasing number of tweeters are tracked, and all these tracked tweeters are considered for both lists. Ranking for the Tweet-1000 is based on average number of tweets per day for the whole period the account has existed. This ensures that account with high numbers of tweets are on top of the charts, but that actually tweeters that tweet a lot are on top of the Tweet-1000.

One thing we learned very quickly is that there is a big difference between interactive human tweeters and non-interactive, often non-human tweeters. Because they are so different, they deserve separate Tweet lists. This will be realized in the very near future.

If you have any comments, complaints, suggestions, ideas or other remarks about Twopcharts please don't hesitate to e-mail me at info@twopcharts.com, or just send me a message on twitter.