How to stop cheating indicators

Hello.

To date, resource developers have a large number of problems. These include the choice of rating system, the method of registration, the fight against spam, the elimination of unwanted comments and content, and many others. Today I want to talk about some ways to combat cheating. I was encouraged by Nikolay Korobko ’s comment to one of my past articles .

For a start, it is worthwhile to decide: what to count as cheating and with the help of what these cheating occur. What can cheat? Rating anything - articles, people, other content and attendance (it makes sense to stop it, if some important parameters depend on attendance).
')

Spin the rating

What is the rating rating? There are two ways - either virtuals are created, or with the help of friends. The methods of struggle are somewhat similar, but there are differences.

Virtuals

Suppose a person creates a virtual. For a more or less significant increase in the rating, he, depending on the resource, needs to create a dozen or more false accounts.

The first barrier should be a unique e-mail. Email must be confirmed by entering a special code sent to a new user in order to prevent registrations of the form hgkh@kljh.ru. But it is easy to manage, free mail hosting is a dime a dozen (this is by the way in their garden, a stone, now it’s easier to tweak new mailboxes easy, I don’t know how to fight, but we have high technologies, thank God). Very elegantly in due time with this problem cost on dforum.ru. There it was possible to register with a unique mail, that is, either your personal hosting, or from the mail issued by the provider. Very clever, but for large portals designed for a wide audience, hardly suitable.

The next bastion is IP. But he is also unreliable in the event that an unscrupulous user worries about his change. And even worse, not all IPs are static.

Next to the IP filter is a filter set by the cookie. But it can be bypassed in two simple ways, or simply clean the cookies, or use different browsers (of course you can hardly collect 10 browser registrations, but for a little hooliganism it will be enough).

In the method with IP and cookies, there is one exception - some people are sitting from the same computer, or they are colleagues, either living together, or sitting at all in an Internet cafe. So here it is not necessary to act categorically. Himself a couple of times ran into such a rake.

And, finally, perhaps the most difficult, but effective way to combat the cheat rating is to follow the assessments that a certain user puts. If the ratings are polls positive or only one category of content is evaluated (comments, articles, etc. from one author, the same series of articles, etc.), then you should take a closer look at the suspicious user. Of course, it is possible to live a double life, but hardly anyone will be able to create a legend to dozens of reg.

"Gangs"

Fighting gangs is more difficult and interesting. You need to be able to analyze a huge data stream, isolating groups that vote strictly FOR each other. However, this also applies to the fight against virtuals.

Another way is to follow the mail. To read it or not to read is the moral choice of every administrator. But the flow of private messages is worth analyzing. Any contact by internal mail is unambiguously regarded as an acquaintance of two respondents. And if you are familiar, it means that they are potentially dangerous :), it means that their mutual evaluations should be analyzed first.

Spinning attendance

You'd be surprised, why stop this attendance? It is a direct road to traffic and shows!

No one is going to force users to visit certain pages more than once. There are just cases when the number of visits to a particular page affects something important. For example, you pay your users for being popular. Or a rating depends on the number of views. Or some other exotic options.

Exit here suggests itself. First, do not count virtual visitors, whose calculations we have already learned.

Secondly, to stop counting visits from unregistered users, people, for example, can simply come from search engines, and this is a random entry.

Third, take into account only the first visit of registered users, or take into account these visits only after a certain period of time, it will be five minutes or a day - you decide.

Do it wisely

Instead of concluding, I want to say that any of the described methods of struggle should not prevent the user from registering and communicating normally. Any prohibition carries a lot of problems. And sometimes you should close your eyes to petty pranks and dirty tricks for the sake of big business, traffic and as a result of the ruble.

Be vigilant and wise!