CareerCup

we can use the "no captcha" technique.. in which humans cant see any captcha with naked eyes (invisible), so proceed with the normal procedure, while the computer detect a captcha n thus getting caught/identified as non human

A 3 step strategy can be used:
1> Real time filter on the date, time and range of IP addresses from where clicks to a ad link are originating
2> Offline analysis of the click ad log files over past days comparing the number of ad clicks for a particular ad link from range of IP addresses
3> Human and machine investigation into false ad clicks involving the company paying for the ad considering data from other ad networks, market conditions, etc.. as required.

click postition within the button.
human click ad button more than once usually in different position, while robot clicks usually in the same position.

why would someone click on an ad if there is captcha...already hardly anyone clicks on ads.....its bad for business....

i said this and immediately took back my words citing the same reasons...interviewer agreed....

Will monitioring the number of clicks from an IP be feasible? It might be more of analytics rather than run time check.

there can be a online filter which inspects the traffic for the ads. The ads must not be clicked with much high rate and also there must be a peek time for the ads click for a particular region. Also not most of the clicks should come from same IP.

1. Min time interval between clicks
2. From which ip/area(router) it is coming
3. max num of clicks in a given interval (~ 1 min)
4. If time intervals between clicks is regular(ex: equal could mean a bot)
5.

Test user agent. Obviously CURL and HTTPUnit are pretty good tells that it is a machine, as is repeated requests from the same IP.

But user agent and IP (via Tor) can be spoofed. So you'll need more. Look at outliars from your empirical data to zone in on potential false positives.

Test user agent. Obviously CURL and HTTPUnit are pretty good tells that it is a machine, as is repeated requests from the same IP.

But user agent and IP (via Tor) can be spoofed. So you'll need more. Look at outliars from your empirical data to zone in on potential false positives.

As far as I understand the question, the question is about how would you identify if the ad being clicked is by a genuine user or a script / virus.
Now for "this" question , I do not understand some of the solutions mentioned here.
1) The check for IP address is incorrect solution as for both the cases user/script the ip address would remain the same
2) People suggesting here to use some kind of captcha, click on this part of the ad, hold left mouse key for x seconds, press various key combinations : guys, the question is about how to "differentiate" between users and script from the click requests that have already come in, and not to "avoid" scripts.

Now I did like this solution : Look at which part of the ad is being clicked continuously by this IP. humans tend to click at different places.

But, to me the questions seems unanswerable as , if the script writer is intelligent enough, he would obviously know that his script should place clicks NOT at constant intervals but at random times and at random places of the ad.

So, this questions highly depends on what the interviews wants to ask you, assume script writer is not much intelligent and differentiate between clicks that have come already.
OR avoid scripts.

As far as I understand the question, the question is about how would you identify if the ad being clicked is by a genuine user or a script / virus.
Now for "this" question , I do not understand some of the solutions mentioned here.
1) The check for IP address is incorrect solution as for both the cases user/script the ip address would remain the same
2) People suggesting here to use some kind of captcha, click on this part of the ad, hold left mouse key for x seconds, press various key combinations : guys, the question is about how to "differentiate" between users and script from the click requests that have already come in, and not to "avoid" scripts.

Now I did like this solution : Look at which part of the ad is being clicked continuously by this IP. humans tend to click at different places.

But, to me the questions seems unanswerable as , if the script writer is intelligent enough, he would obviously know that his script should place clicks NOT at constant intervals but at random times and at random places of the ad.

So, this questions highly depends on what the interviews wants to ask you, assume script writer is not much intelligent and differentiate between clicks that have come already.
OR avoid scripts.

Probably before redirecting to actual add website ..use some sort of captcha.This captcha would come into picture once user clicks on add.