5 ms·
DoubleClick's ad domains are hard coded into Google Chrome [1], so that additional telemetry is sent to their ad servers on every request. You cannot disable it
by 0xy 5y ago
DoubleClick's ad domains are hard coded into Google Chrome [1], so that additional telemetry is sent to their ad servers on every request. You cannot disable it. It's never disclosed to users.
When you use Chrome, you're literally using the DoubleClick Browser.
Privacy and ethical concerns aside, this is a clear antitrust violation. Google's ad competitors are at an automatic disadvantage.
The FTC has gone after Intel for releasing a compiler that deliberately disadvantaged competitors, but Google doing much more egregious evil stuff to ad competitors isn't touched. I guess it pays to have the House of Reps Speaker Pelosi own copious amounts of profitable call options in Google. [2] It's profitable to look the other way, isn't it?
[1] https://chromium.googlesource.com/chromium/src/+/e51dcb0c148a1b0ab0080808d8d1b914d02f6057/chrome/browser/metrics/variations/variations_http_header_provider.cc#35 https://chromium.googlesource.com/chromium/src/+/e51dcb0c148...
[2] https://nypost.com/2021/12/30/nancy-pelosi-buys-call-options-for-google-disney-stocks/ https://nypost.com/2021/12/30/nancy-pelosi-buys-call-options...
- tremon 5y agoyour [1] link references a comment in the code, not actual working code. Can you do better or should we classify your entire post as fake news?
- w1nk 5y agoThis comment seems unnecessarily aggressive given the code is pretty straightforward to navigate. If you were to read the source code a little further you'd jump into this code, called from OPs referenced function: https://chromium.googlesource.com/chromium/src/+/e51dcb0c148a1b0ab0080808d8d1b914d02f6057/chrome/browser/metrics/variations/variations_http_header_provider.cc#220 https://chromium.googlesource.com/chromium/src/+/e51dcb0c148... I'm not sure exactly what kind of 'proof' you're looking for , but this tracking isn't being hidden.
- tremon 5y agoThanks. The OP comment refers to line 35, while the actual code starts at line 232. I still think it's unreasonable to expect someone to scan 200 lines of source code to validate a claim.