> I don't really wanna' say what his methods were since it will kinda' give away some
> of the things our safeguards protect against, but it became pretty evident after
> looking at a few of his entries he wasn't just making honest mistakes.
As I told to somebody by PM, the job is done so fast that we can discover messes like these only after the job has been completed. Luckuly enough, we had forseen that somebody might have tried to mess up so we setup a safety routine preventing people to access other people's work when marked as completed. This prevents sabotage like somebody deleting or overwriting with trash other people's work. If we hadn't this, the damage might have been MUCH worst...
For the future, my plan is to setup a compare routine with functions like php's levenshtein() to detect when inputed text is TOO different from other people's entries (some bits may vary, but if 20% is different there is something odd going on!) and mail-report us, so that we can try to block some users by banning IPs, or pruning the sabotage.
The system is anyway growing, each run includes new features and all is studied so that things are modular and can be just tuned and reused (see the mutiple image handling for this run, the system self detects if there is a alternate image and puts the link if applicable ). I just wish I had some more spare time to setup things in anticipation instead of running to cover holes...
The first Italian MAME resource at
Edited by s_bastian (03/23/11 01:33 PM)