Clicky

What is "bots"?





What is bots?


From time to time encountering the bots, we are used to treat them as a rubbish that pollutes the social networks. We perceive them an unavoidable evil, as a part of reality that we have to put up with. Sometimes it seems that bots have appeared at the same time with social networks.


But we forget that bots can’t create themselves, they are registered and operated by people. Only by knowing the extent of the problem is possible to estimate the real damage caused by its existence. Number of bots in social networks is hundreds of millions, it doesn't look like a harmless game.


If we consider the bots not as technical abstraction, but as a special tool which is purchased to simulate the audience on the media market, their classification becomes logical and clear.


What products on social media market have the greatest demand (on any platform)?


1. Followers (friends, subscribers)
2. Retweets (shares) and likes
3. Trending (what's popular right now)
4. Traffic (visitors from social media)
5. Complex campaign on social networks with a lot of metrics (as a part of major media campaign)


Why do people buy these services? Their objectives are obvious, they need to simulate the size of its audience. Vanity and pride are not the most important reasons for this. People deliberately overestimate the scale the audience or imitate the entire audience in order to profit.


There are 3 types of bots that have similar characteristics and perform the same functions. We consider them mainly by the example of Twitter, but this typology can be used in the study of any other social network.



1 level bots.


Passive (static) audience, the general popularity rating. Primitive simple followers

What is most important to this type of bots? There should be a lot, and they have to be cheap. Their main features are based on just these two parameters.


They have a lot following, this is the main task, for which their creators get paid. Because these bots need a lot, and the price should be very low, it is not profitable to spend time on their thoughtful design. It usually has a small amount of tweets and followers, no description/location and default picture. Sometimes it even has "some kinda description" and a random picture, to have a little humanlike preview. But on closer examination it turns out to be primitive bot.


First level bots are not intended to a credible imitation of user. Their creators just sell cheap low-quality audience, the only indicator of which is the number of followers. Another reason for their poor quality, that it is impossible to verify millions of accounts with your own eyes. Even if you check 10,000 (and this is a very large number), it is still a test of just 1% of the million.


For a more identification of bots 1 level, we have developed evaluation system STAQ (Simple Test of Audience’s Quality). It includes the assessment of 12 parameters to determine bots 1 level with high accuracy.


Fake popularity is one of the most effective technique to attract real audience. When you see two accounts with the same content, but one has a thousand followers, and the other has a million followers, you sign up for the one that is already popular.


The average percentage of bots this type among followers of Twitter Top 100 is 58%.


The total number of bots 1 level amounts to hundreds of millions.



2 level bots.


Active (dynamic, valid) audience. Retweeters, likers, hashtag- and link-posters.


Unlike level 1 bots that simulate a wide audience that has ever interested in this account, the level 2 bots imitate situational attention to certain tweet or event. These type of bots perform repetitive periodic actions, imitating the audience’s interest in a particular topic, statement, subject. Level 2 bots simulate activity of the audience that allows you to show the growth of supporters / fans, and to explain the sudden viral rise in popularity.


What are the indicators of the number of passionate users? Firstly, it should be retweets and likes. Second feature is the trending of hashtag at the regional, national or global level. The third one is the number of links to articles on the topic. We remind you that the high citation index in social networks improves the relevance of the article in the SERP (Search engine results page).


So, bots 2 levels can be distinguished 4 specific type of content.


1.Retweets (shares, reposts).
2. Likes.
3. Hashtags.
4. Links.


This type of bots are usually more advanced, than level 1. Looking at them, it could be hard to define, it is a bot or a human. The level of masking can be very high. Profile pictures can be carefully selected, retweets are collected by certain topic, the number of following isn’t abnormally high, it correlates with number of followers.


But they have some statistical anomalies, which shows the purpose of their creation. Their content is often very monotonous, the account has an excessive amount of retweets and likes, or hashtags and links or mix of all kinds of activity. Some of them a very high rate of tweets. Quite often, the second level of the bots mimic the action of the youth audience, teenagers or students.


Sometimes we can find also disposable level 2 bots. Their quality is similar to the level 1 bots, they are active just one to two weeks after join the Twitter, then they are moved to the reserve until the next occasion.


To determine the level 2 bots used content analysis. The easiest way to determine this account is high number of retweets, likes, repetitive hashtags or links.


The total number of bots 2 level amounts to tens of millions.




3 Level.


High quality target audience. High-quality imitation communities and personalities.


3 level bots (B3) is the high-quality imitation of common users. Their stats looks like human, it has no anomalies to detect them, like extra tweeting or mass following. Its production usually takes several years, during which they regularly tweet something in accordance with professional or personal interests. These accounts can have a several photos of the “owner”, a credible identity, and its content is very similar to human.


If they have so much in common with humans, maybe they are real people?

No, unfortunately that is exactly bots, despite the best efforts of their creators. Although their actions looks very similar to human, communities consists entirely of bots. There is no communication between real people and bots, these circles don’t intersect.


In a simplified way, people don’t follow bots. It is absolutely meaningless for people. If a real person follows the bot, whether it can be called a social network?


Well, maybe some kind of people follow the bots? But if you believe that the following bots is really stupid, why other people should do this? In social networks people follow only accounts with interesting content, or their off-line friends.


Of course, there are some exceptions. This is primarily mass followers and active back followers, who follows tens or hundred thousands accounts, most of which are bots.


In fact, the mass back following is often a form of statistical fraud. Actually, this is one way of additional verification of bots as a humans, when they have popular or verified accounts in followers. Moreover, it is an excellent protection against the suspension.


Level 3 bots are used as the most realistic targeted followers, sometimes as extra targeted audience in expensive ad campaigns for some humanlike actions. Most of the time they mimic moderately active average user.


Among these bots can find various simulated professional groups, such as programmers, lawyers, traders, journalists, marketing, SEO and SMM specialists, and many others.


It is important to add, that to create such bot communities (not a primitive botnet), you need to have not just programming skills, but professional knowledge in sociology and marketing.


The total number of bots Level 3 counted in millions.



Other types of non-human accounts. Cyborgs


Ordinary bots, often level 2, on whose behalf the person writes. Usually such a person runs several tens or even hundreds of such accounts. Often it acts as a Troll. Most importantly, any real person inside a botnet can’t be classified as a human user, because its sole purpose is to simulate a human behavior of a false audience.


The blanks (sleeping bots)


It is temporarily inactive accounts that are in standby mode after joining Twitter, or after a previous use. The amount of such accounts is comparable to the number of level 1 bots. Typically, these accounts are used for the creation of cyborgs or trolls. According to the date of registration, they are perceived as reliable old users, although active was only the recent period.



Summary. The user is bot, if...


If it has some statistical anomalies.

If its main purpose is to increase not its own, but someone's popularity.

If it follows bots, or if it is followed by the bots, except the mass followers.


Note 1.

These three conditions can be used in any combination or each separately for different type of bots. Items are listed in ascending order of importance.


Note 2.

This basic classification covers the vast majority of bots. The use of this technique requires some research experience. In some case studies may be found subtypes or mixed types of bots or other manifestations of artificial audience.