[chatmasta]

But advertisers want to target their advertisements to specific demographics. How can you target the user (identify the user's demographic, to select which ad to show) if you can only collect limited information based on the load of the static image?

Ultimately you would end up with a model where advertisers target based on website, rather than user demographic.

[jdsin]

Good question. Maybe target using some kind of heuristics? We can figure out who is loading our imgs: ip addresses, requesting browser details, load time, approximate location (city, state, country), what website is the user visiting (can figure out what they are interested it) etc. We can build anonymous profiles of users. Ofcourse, the bigger the network, more precise the heuristics..

[chatmasta]

Right. So you've just arrived back at the status quo of tracking!

How do you define "anonymous?" Even if you have a policy not to store "personal details" like real name, you are still tracking some set of attributes of each user (ip address, headers, latency, etc). Together, a set of these attributes is likely sufficient to uniquely identify the user the set belongs to.

So now you're back to the original problem.

[jdsin]

Dang it! Back to square one. So it seems its a double edged sword. Personally I feel this is a good compromise. What would you rather have, something like this? Or something like Google Ads that follow you throughout the internet with help of super cookies and javascript, merge everything they know about your behaviours from their other services, and serve ultra specific creepy ads?