posted on Sep, 24 2009 @ 11:58 PM
Seems like a good place to place this so...
I have been noticing a lot of buzz on ATS and other places about the emergence of the massive data monitoring and data mining projects in the US and
Europe that will sift through all the data that gets collected on you to look for suspicious patterns that might indicate a possibility that you might
do something someday. Of course this is all wrapped in one of the two standard reasons of either keeping us safe from terrorists or hunting down
paedophiles.
Historically we know that any time something this Orwellian is introduced for either of those reasons, it soon becomes too tempting to use only for
that. For example, every nation that has instituted internet filtering for “child porn” winds up filtering our sites critical of the government or
that it just doesn't like. Just consider the massive abuse of the Patriot act to regularly circumvent normal civil rights when it is just to darn
inconvenient to follow them while doing police work.
The simple fact of the matter is that whether we like it or not, these massive data collection systems are on line and collecting data about you and
there is not a thing you can do about it.
Of is there?
You see, this data only means something if it is representative of your normal, everyday activities. In other words, the data mining works backwards
from the data using patterns it finds to reconstruct what you are doing out there in the real world. But that only works if the data is really
representative of what you do. See where I'm going with this?
First of all, the terrorists and paedophiles and all those other people that are supposed to get caught, at least the ones that really do all that
stuff and not some moronic wanna-be's that only get caught because some FBI undercover agent patiently led them through the basics; not them but the
those real pros -- the ones trying for the suitcase nukes -- the guys we need to catch already know how to not get caught by this big brother
database.
For example, the TSA is looking for patterns of ticket purchase and seat selection to identify terrorists. They are obviously also using ethnicity
and other factors but won't say so for political correctness. If I'm a terrorist, then I'm going to change my apparent behaviour by creating data
that looks like someone who does not fit the profile. I change my name to an Irish one, I become a life-long baptist and republican, I create a
regular pattern of travel between Dallas and Washington over a multi-year period using a credit card to pay and always asking for a window seat.
Things like watch lists only work when those being watched are unaware.
We need to poison the well (so to speak) of these big brother systems. To reclaim our privacy, we need to make the reliability of the predictions and
usefulness of these massive databases and mining models so low that they are not worth doing any more. What about finding those real terrorist guys?
Isn't this going to make America/Europe less safe? No. But more on that later on so hang tight.
How do you poison the well? We have to do it in significant numbers to make an impact. Here are some things we can do.
1.Use cash. Worked for our parents and grandparents. Why do you have to put dinner on your credit card? Buy gas with your bank card? This means that
the more you work off line, the more incomplete your data is.
2.Contra purchase: that means I buy stuff for you and you buy stuff for me in ways that leave records. It works best if I'm a 65 year old male with
prostate trouble and you are a 22 year old pregnant woman with a small family. My data now represents two different people/families and their buying
patterns and not just me. Now extend that to contra buying with a group of 20 people. Sort that out!
3.Random activities. Regularly do bizarre internet searches (goggle sailing a lot if you live in Kansas), register on knitting websites, take out
library books written in Swedish, join all the political parties in your community. Get everyone else to do that too. Populate those databases with
all kinds of bogus data. Drive around aimlessly and wave at all the CCTVs.
Once we start to deliberately create data that is not representative, the underlying assumption of most of these predictive data mining techniques no
longer holds and it is all rubbish.
Oh, but wait, what about catching real terrorists? The big brother approach doesn't work. Anyone who works with this stuff knows that. What does
work is good, old fashioned human intelligence assets out in the field doing the kind of intelligence work that has been done for decades. If you
follow the news, most of the terrorist arrests have come from exactly that – good field intelligence work that is supplemented by TARGETED data
collection on specific individuals.
Should we use data mining? Yes, but not as a tool for sifting through the general population – the potential for abuse is too high. But a small
group that can use whatever means necessary to hunt down real terrorists is the way to go, but only with stringent oversight to ensure that they stay
on mission and do not become tools for the political masters to weed out “wrong thinkers”