There is no single way to guarantee the uniqueness of a user on the Internet.
It is impossible to determine the difference between the bot and the real user.
Taking into account current trends in computer vision and training to prevent the use of captcha, it may seem that captcha is no longer an effective tool for identifying bots.
For any PAT (port address translation) combined with NAT, the network structure hides internal computers. There may be variants of USERAGENT indicating unique computers inside (provided that they were not fake).
But a set of twenty computers cloned from the same OS source and for PAT will look almost identical, the ports will be different, given the randomization of source port mappings and when compiling using PAT translations, the PHP Internet web server will still be displayed as one computer.
Regardless of whether the user is the same, the information cannot be set:
Cookies only identify specific computers associated with specific browsers and specific users. If the user would like to triple or quadruple ... himself, he can launch different browsers (IE, Safari, FireFox, Chrome and Opera) or several logins on the same computer (not to mention curl, wget and many others). access methods). It is still not able to distinguish individual users posing as several.
There are other network factors, such as Tor exit points and proxies, which further complicate the paradigm.
Considering that people tend to the same viewing patterns, given enough data, this can be set as the probability that users are the same, but even this probability can never be reliable (not to mention random bot patterns).
Logging in has a greater effect on the “user problem”, if you require each user to have a verified email account, you are more likely to have one user to log in, especially if you exclude common public mail domains such as hotmail .com and gmail.com and more. Private domains are nearly impossible to validate by individual users. The percentage of trust increases with a well established / well-known private domain (e.g. mit.edu).
In conclusion, this goal of identifying a single user cannot be reasonably achieved with any certainty from clients over the Internet .
There are mitigation methods available, but none are reliable.