Comment by MichaelCrawford | Hacker News Reader

Comment by MichaelCrawford | Hacker News Reader

Parent Full thread

MichaelCrawford

The easiest way to spot web bugs is to use a very old build of Safari. There are many other ways but if you have an old Safari select Window -> Activity then leave it open as you browse a few different domains.

You will see some 43 byte documents with huge long URLs full of query parameters, also one byte javascript sources. I block the ones I find in my hosts file:

   127.0.0.1 www.hosted-pixel.com

On some operating systems it may be better to use 0.0.0.0 but I am not completely clear. Alternatively block them in your firewall.

Among the reasons I dont install any other mobile apps other than those I absolutely require - not even free ones - is mobile analytics. The developer SDKs are all free as in beer but I have seen a photo of one of their data centers. Data centers are expensive; someone must be paying for all that.

You can edit the hosts file on iOS with iFile from the Cydia app store if you jailbreak, alternatively you can maintain the hosts file on your box then install it on your device with scp.

Similarly for Android but I dont know what text editors you can use to edit system files. Some Android devices enable you to install your own firmware build but I dont have a current list.

I am concerned about the impact analytics may be having on democracy. Its not really a secret ballot if the candidates all know what pages I read.

"Among the reasons I dont install any other mobile apps other than those I absolutely require - not even free ones"

This. People are up in arms over the web. If they only knew what each app in their pocket was sending.

Tool to maintain /etc/hosts: https://github.com/jakeogh/dnsmasq-blacklist

Just last night it occurred to me to write browser add-ons that would scramble those query parameters, also send bogus user-agent headers but only for the web bugs.

That is, the add-on would discover one-pixel transparent gifs, take note of what query parameters they used, then every time it found that same gif in the future it would issue a GET with randomly selected parameters drawn from the instances of that same gif it had seen in the past.

For extra crispy electronic warfare, that same add-on could issue GET requests at randomly selected intervals. If those intervals were reasonably far apart (say one hour apart for any one gif) then it would not obviously be an attack on the analytics server.

What it would do is to make it completely useless to correlate your visits to different domains.

Why bother randomizing the query parameters, when you could just permanently cache the image?

If I permanently cache the image or block the server with my hosts file then the web analytics services will not track me.

If I randomize the query parameters then they will not track anyone because they won't provide useful information to those who presently purchase it. Of course that will require far more people then me to also randomize their query parameters.

Reply on news.ycombinator.com