[RESOLVED] Optimal method to prevent visitors voting twice or counted twice?

Confessions

I know the method will involve reading and logging their IP / username, but wouldn't it require quite a lot of space for bigger sites like Youtube? Do they just store a huge array of IPs / users who have viewed or voted on a specific video? Then whenever a person votes, they would pull out that array and do matches.

Would it slow down your site if you had these arrays of up to 1000 IP addresses (having to check through those arrays whenever a visitor voting)? Do you store them just as they are, or ip2long() them?

laserlight

If you require registration to vote, then recording the user id of the voter is an obvious choice.

Confessions wrote:
Would it slow down your site if you had these arrays of up to 1000 IP addresses (having to check through those arrays whenever a visitor voting)? Do you store them just as they are, or ip2long() them?

Store them in a database. Storing an integer instead of a string is likely to be better, but it may not matter when indices are used.

djjjozsi

in youtube when you vote, there (is a)/(there are) connector table(s) to store the videoID, userID, vote_value, date_time values

if you set a UNIQUE index on videoID and userID pairs, on insert that eventuates an error code (mysql_errno() will be 1062)

...did not filter the users by IP address, because in a student hostel or in a company network all the visitors outgoing IP address will be in a short range, so you should not count on IP address to identify 'a' user.

laserlight

djjjozsi wrote:
...did not filter the users by IP address
, because in a student hostel or in a company network all the visitors outgoing IP address will be in a short range, so you should not count on IP address to identify 'a' user.

Oh yes, I forgot to add that: in Singapore, we have a rather limited range of IP addresses, so I get rather frustrated when people use IP blocking.

Confessions

I see. Yeah of course you wouldn't identify registered users using IP addresses, but some sites don't really have many registered users on them, such as those public newspaper sites whenever they run a poll.

So basically just store them together, with a special delimiter. Check through this database field whenever someone tries to vote, to determine if he or she has voted before.

Thanks for the help!!

PS. I was just a bit surprised, when you see some Youtube videos having over 100 million views. Especially since visitors used to and probably still do, contribute to the 'total views' and not only registered members.

laserlight

Confessions wrote:
So basically just store them together, with a special delimiter.

No, store them in rows, and thus obtain advantage from the use of indices. Avoid multi-valued fields.

Confessions

laserlight;10949452 wrote:
No, store them in rows, and thus obtain advantage from the use of indices. Avoid multi-valued fields.

Thanks a lot for bringing that up! At first I wasn't sure what you meant, so I observed how phpBB stores their poll votes in the database every row.

Would it be safe to detect IP addresses this way?

$ip = $_SERVER['REMOTE_ADDR']

I know proxies and stuff will work against this, but as long as no one can bypass that easily and send in 10000 votes within an hour it would be ok.

laserlight

Confessions wrote:
Thanks a lot for bringing that up! At first I wasn't sure what you meant, so I observed how phpBB stores their poll votes in the database every row.

For further reading: Database Normalization And Design Techniques 🙂

Confessions wrote:
Would it be safe to detect IP addresses this way?

$ip = $_SERVER['REMOTE_ADDR']

I know proxies and stuff will work against this, but as long as no one can bypass that easily and send in 10000 votes within an hour it would be ok.

I guess that depends on your notion of "easily", but it might be okay.

bradgrafelman

laserlight wrote:
I guess that depends on your notion of "easily", but it might be okay.

For example, do a Google search of 'PHProxy' and you'll get over 500,000 hits. It's probably a safe bet to assume at least 1/5 of those are the actual proxy or at least link to it, so that's about 100,000 different IP's I have at my disposal to use for voting. 🙂