Standard Mischief

Just a few days ago I told you about how AOL accidentally released a bunch of user’s Internet search term records, which was exactly why I had suggested way back in January that everyone zero out their Google Cookie.

So there is a new plugin for Firefox called TrackMeNot, and it sounds like a good idea:

TrackMeNot runs in Firefox as a low-priority background process that periodically issues randomized search-queries to popular search engines, e.g., AOL, Yahoo!, Google, and MSN. It hides users’ actual search trails in a cloud of ‘ghost’ queries, significantly increasing the difficulty of aggregating such data into accurate or identifying user profiles. TrackMeNot integrates into the Firefox ‘Tools’ menu and includes a variety of user-configurable options.

Of course I’ve been trying it out for a few days. Here’s a selection of my true search terms co-mingled with the TrackMeNot ones.

(More below the fold)

Can you pick out the fake ones?

chokes hats
2000 feet in miles
navigatable waterways
condoms Microsoft
navagatable waterways
kit frogging
fair use single photos
CSS pseudo-elements
?pseudo-class? lang
pseudo-class is :lang
stylesheet highlight text
emergency spoke tool
accumulitis
diffing cretinous
lints uploaded
lottery virginia consortium
warts quarters
lottery virginia legitimate agent
thrashing servers
lottery virginia ?legitimate agent?
diffing cretinous
lottery virginia legitimate
robots Usenet
lottery virginia cover
tooling ken
lottery virginia cover
mig wire
freedom of the reloading press
faulty munches
?freedom of the? reloading press
swizzle mailbombed
pistol virtual
cretins ice
cretinous plonking
fossil bums
lincon welders
lints uploaded
lincoln welders
?spyware, in my computer? ?more common than you think?
inurl:kb mozilla firefox history
elegant pushes
inurl:kb mozilla firefox
depredation permit
tracfone 10000..99999 expires
philtap
booting confusers
nybbles
spoof flaky

I don’t know how hard that was, I’m already in the know. but I think those smart cookies over at Google can figure stuff like this out pretty quickly. Here’s the fakers from TrackMeNot:

swizzle mailbombed
pistol virtual
cretins ice
fossil bums
thrashing servers
spoof flaky
kit frogging
diffing cretinous
lints uploaded
cretinous plonking
chokes hats
tooling ken
condoms Microsoft
elegant pushes
booting confusers
warts quarters
faulty munches
robots Usenet
randoms softy
wallpaper nuked
obscure nanotechnologies
nickle jiffy
chain clobbers
kicks pistol
muttering NeWS
cubinged bumped
lamer foregrounds

Notice anything? they are all two word queries. Furthermore, when I was preparing this post I found this on Boing Boing:

Odiumjunkie- As it stands, the extension creates “random” search terms by combining words from a (very short) wordlist in a pseudorandom manner. The list can be seen in the extension’s source code, and I also put it up here; http://tinyurl.com/f7n8u . It contains around fifteen hundred words – not nearly enough for the intended obfuscation to be effective, as it would be trivially easy for any party with access to the data to screen out search entries consisting only of those words.

In contrast, here were my real searches (in bold). Everything including and after the # is a comment:

2000 feet in miles #google used as a units converter
navigatable waterways
navagatable waterways #a quick refining of the spelling on this two word query
fair use single photos # four term search
CSS pseudo-elements # the dash gives it away
?pseudo-class? lang
pseudo-class is :lang # again, evidence of refining the search, non-letter text, and more than 2 queries
stylesheet highlight text
emergency spoke tool
accumulitis #again, single term
lottery virginia consortium
lottery virginia legitimate agent
lottery virginia ?legitimate agent?
lottery virginia legitimate
lottery virginia cover
lottery virginia cover # refining a search combined with requesting results 11-20
mig wire # wow, the first two query real result
freedom of the reloading press
?freedom of the? reloading press #Quotes, refined search, five terms
lincon welders
lincoln welders # two word query, but the first search was misspelled, If I remember correctly, Google suggested the spelling ?lincoln?
?spyware, in my computer? ?more common than you think? #Duh
inurl:kb mozilla firefox history
inurl:kb mozilla firefox #refined search, advanced google options, etc
depredation permit #second case of plain-jane two word query
tracfone 10000..99999 expires #advanced search query, etc.
philtap
nybbles #single term followed by a dictionary lookup.

So what’s the result here? I’d have to say nice idea, but this is not ready for prime time. It’s pretty easy to ferret out most of my real queries, and that’s before we’ve even checked those two, 2-word queries against the wordlist. This thingy needs a lot more code before it might even start to look like a real person.

Related Tags:
TrackMeNot, Google, AOL, Firefox, firefox extension, Search Querry, Search terms, obfuscation, obfuscate