[Home | Contact | What's New? | Products | Services | Tips | Mike |
Living with Schizoaffective Disorder

Solving the Analytics Problem

A Domain Name System server would return for servers that serve single-pixel transparent gifs.

Jonathan Swift

Safari used to have a very useful Activity window, that displayed the size of each resource that composed a web page.

The minimum size for a single-pixel GIF is 43 bytes. So I'd keep the activity window during my day, then from time to time I'd look up the domain names of the servers that served 43-bit resources.

Then I would blackhole them in my hosts file: www.hosted-pixel.com # I Am Absolutely Serious

The worst offenders were political campaign sites.

Analytics services are often free of charge to the webmaster, but require entire data centers to serve nothing but single pixel transparent gifs. Data centers are expensive - someone must be paying for all that information.

Studies have shown that it is possible to identify individuals from the websites they frequent. For example I mostly use gmail, kr5ddit.com, soylentnews.org, google news and Facebook. I mostly visit them in the late morning and late at night, always from the same Comcast IP address.

I am completely cool with advertising but cannot tolerate tracking. The simple contemplation of how much I am being tracked leads my paranoia to flare up. Enough of that I would hallucinate too.

Consider the challenges faced by closeted gay right-wing politicians.

This could present a solution to the problem of ad blockers. About 30% of web users employ an ad blocker. This has resulted in plummeting revenue for ad-supported sites, most of which offer their content completely free of charge.

The use of ad servers, in which each advertisement is fetched from a domain different than that used to serve the content, enables tracking as well. Consider google Adsense. Webmasters add a snippet of javascript to their pages that fetches the markup of each ad from google's server. Adsense is popular, so google's Adsense servers quite likely perform analytics tracking as well.

Ads served by an ad server are easily blocked by ad blockers such as ublock origin, which can see that a resource is being fetched from a domain other than that used to serve the content.

But there is a simple solution solution that overcomes ad blockers while defeating analytics: serve the advertisements from the same server as used to serve the content. While not commonly used, there are some ad services that provide a back-end API that your contain server uses to fetch the ads.

But all the ad servers see is lots of GET requests from the content servers. They are unable to obtain the IP addresses of any individual end-users.

It is easy to edit your hosts file if you are technically inclined. But the vast majority of website users are _not_ technical; even many power users just do not want to go to the trouble.

Another problem is locked-down operating systems. While iOS (iPhone and iPad) has a hosts file, one must jailbreak one's device to edit it. While jailbreaks are widely available, again most people are not technical enough to avail themselves of them. There is also the risks that firmware upgrades will brick jailbroken devices.

There is a very simple solution to this, that is far easier for end-users that editing hosts files: enter custom domain name system servers in one's network settings. This will work on all devices that I know about.

Even non-technical users would be readily able to follow well-written instructions.

There are already some open source domain resolver packages that perform black holing, mostly for use in defeating open spam relays.

To collect the list of analytics servers I would write a javascript add-on for chrome and Firefox, and a native plug-in for internet explorer and safari - the latter do not support add-ons.

This javascript would identify web bug servers then save their hostnames to a text file.

I would enlist some volunteers - I do not need many - to use the add-ons or plug-ins for a month or so, then they would upload the log files to a server.

The files would be combined to produce a blacklist, which would be supplied to a suitable domain resolver.

There are ways this could be done that require payment by our end-users, but the authentication would be such a hassle for them that few would go to the trouble.

I propose to monetize the service by seeking donations. This kind of thing would work really well for a crowdfunding campaign. The service's website would have a donations page, that would contain wallet addresses for cryptocurrency, as well as an Amazon Honor System tipjar and a PayPal donate button.

The analytics vendors would howl like wounded hyenas. I would politely invite them to eat my shorts.

[Home | Contact | What's New? | Products | Services | Tips | Mike]