4 ms·

Sounds like a nice solution. It does not sound trivial to me, though. Do you have any code to share or other pointers?Especially around the 'tuning' of this sys

by bestes 18y ago

Sounds like a nice solution. It does not sound trivial to me, though. Do you have any code to share or other pointers?Especially around the 'tuning' of this system (like, how many hours was too many, how much distance was OK, etc.)

buro9 18y ago
It was trivial to collect the information. First you start with what the server knows, and log each data point. Then what can JavaScript tell us? Log each data point. What can common plugins tell us? Windows Media player has a UID cookie, call it twice from different IP's and see if we get the same number back. Flash could tell us a bit more too. Then there are the geospatial properties, where, when? Log those. We logged everything we could, and then pushed them over to an MIS layer. The MIS layer included views that ranked all users according to various basic criteria (number of hours when login has been performed per day, total amount of time spent on the system (windows media log files), number of unique machines used to access content). With those views the business could then use existing reporting tools to tweak the criteria or investigate a more severe anomaly in a set of data points. The result reports went back into the system and became the new benchmark for the next set of scheduled de-activations or emails. This way the computers could be left to do it 99% of the time, and people could come in and tweak it visually for the 1% of the time. We were purposefully tolerant of the issue as it is better to let a few people through and gain that marketing and audience than it was to have false positives and the bad marketing on the sport team forums. I can't share the various bits of code I'm afraid, it's a while ago and belongs to my employer at the time, though the technique I've outlined above.
- teej 18y ago
  Thanks for posting this. For a long time, I've had the opposite issue: detecting and disabling single users running multiple accounts. This is a much more comprehensive way I could narrow down where requests originate and then map that to clusters of user accounts.

by bestes 18y ago

buro9 18y ago

It was trivial to collect the information. First you start with what the server knows, and log each data point. Then what can JavaScript tell us? Log each data point. What can common plugins tell us? Windows Media player has a UID cookie, call it twice from different IP's and see if we get the same number back. Flash could tell us a bit more too. Then there are the geospatial properties, where, when? Log those. We logged everything we could, and then pushed them over to an MIS layer. The MIS layer included views that ranked all users according to various basic criteria (number of hours when login has been performed per day, total amount of time spent on the system (windows media log files), number of unique machines used to access content). With those views the business could then use existing reporting tools to tweak the criteria or investigate a more severe anomaly in a set of data points. The result reports went back into the system and became the new benchmark for the next set of scheduled de-activations or emails. This way the computers could be left to do it 99% of the time, and people could come in and tweak it visually for the 1% of the time. We were purposefully tolerant of the issue as it is better to let a few people through and gain that marketing and audience than it was to have false positives and the bad marketing on the sport team forums. I can't share the various bits of code I'm afraid, it's a while ago and belongs to my employer at the time, though the technique I've outlined above.

teej 18y ago
Thanks for posting this. For a long time, I've had the opposite issue: detecting and disabling single users running multiple accounts. This is a much more comprehensive way I could narrow down where requests originate and then map that to clusters of user accounts.