A note on @NotetoSelf

My 100% favorite podcast right now is
Note To Self. Most notably, it forces me to address principles and ethics that initially stirred my interest in programming as a child. As our species spirals ever closer to The Singularity and the infamous post-privacy Orwellian dystopia we’ll bequeath our children, projects like this pull me from the haze of Amazon Prime purchases and the horror of the casually-agreed-to End User License Agreements that govern the devices I use every day.

On Privacy

We’re at an un-precedented time in human history. The vast access to information and the ease in which you can be an active consumer whore, ever deepening in data mass and Customer Lifetime Value, brings the risk of stagnant analytical thought and problem-solving ability to all of society. We’re headed fast down the road to a civilization comprised largely of automatons. The warnings fall, to most, on deaf ears.

Unless you feel a deep sadness growing in your soul each time you see a finely-tuned, borderline creepy digital ad, this may not be an article for you. Go buy some moderately-priced appliances!

The end goal of data mining is not to sell you something. That’s today. That’s already happening. It gets better every day. As questionable as many digital advertising practices are, it’s a far cry from the possible applications of infinite user metrics. The end goal is control.

It begins, these days, with control over your buying patterns. It ends when we are Borg.

Consider a few short exercises in which I use an existing algorithm (set of rules) to serve an average user an ad:

Exercise 1: The manipulation of consumer buying patterns via disposable manufacturing standards combined with scheduled ad tailoring

a. You buy a toaster on a website that tracks consumer activity.

b. The toaster is made with common modern manufacturing standards; it’s essentially disposable, and very affordably priced. This allows the retailer to schedule when they can next serve an ad to you showing this same type of product, so you can buy one again. I know what you’re thinking – “I wouldn’t buy the same toaster!” – it doesn’t matter. There’s more than one brand of toaster on amazon.com. They’re all shown to you using the same algorithm.

c. The toaster you bought breaks in 1 year. You log in to a social media site, and see an ad for a sale on appliances that the algorithm has scheduled for you. It knows your piece of crap toaster is broken.

d. You click the ad. The site algorithm records that this particular targeting has worked, and tailors your ads to be shown in a similar manner in the future.

When you read the example above, it can be easy to dismiss. The dark implications aren’t always apparent to those unfamiliar with thinking in algorithms, even less so, learning algorithms that continuously refine and adjust themselves in response to user data.

Let me go a bit darker with the next example. My apologies if you become offended. Please know that I am not trivializing sexual offenses; I use it to illustrate the depth of depravity and callousness in the data-mining development practices around us.

Example 2: A rape victim with a low-paying job is offered discounted therapy sessions

a. Someone is a victim of a sexual assault.

b.* The victim survives, and is hospitalized. While in the care of the hospital, the victim decides to pull out their phone, and fill in their friends. They thank everyone on their social media account of choice for the support, the hospital visits, etc. One friend, via “private” message, asks for details. The victim of the sexual assault responds, explaining what happened. The hospital wifi connection is verified by the social media site, as well as via GPS coarse geolocation – something that can roughly triangulate your position without you turning on location services.

*A note on step b: This is assuming insurance companies are still prevented from selling patient information at this time. If medical establishments ever allowed to disclose data from your medical operation / condition / care to any advertising corporations, data like this can be gathered directly from hospital / clinic records. Sound insane? I bet you’d agree to it for $2,500 off of your neo-natal care bill.

c. Several words are flagged in the social media messages/comments; perhaps keywords such as “rape”, “cope”, “pain” “need to talk to someone”, “I’m here for you”, “how can I go on”. You get the idea.

d. Given enough keywords being present, along with any other data available (a scanned police report, a check-in at the hospital, a police information phone number posted by the victim’s parents ), this triggers a probability-powered algorithm that shows the victim an ad for local therapy sessions.

e. But wait! Before it serves the ad, the algorithm reads the victim’s home address. Comparing this with public statistical data, it matches the zip code to a low-income community with a high crime rate. In this case, a high-priced private Psychologist might not sell. So instead, the algorithm shows the victim an ad for a low-priced group therapy clinic.

So, how do I know this? Am I a paranoid shut-in? No! I’ve done quite similar things with user data. In a normal array of client work. Collecting user data. Mining ad data from Facebook Ad Center, Google Adwords, and the like. And guess what? I’m just an average web developer.

Our privacy is deeply in trouble, any way you look at it. Use a VPN. Block ads. Reduce time with social media. Clear cookies, clear your cache. Let’s at least delay things while we still have the power.

But seriously, please use a VPN. It's like locking your door, and closing your curtains at home. If you don't do it, you're just kind of inviting sketchy people to take a peek.