The Problem with Algorithms

I’m reminded on a daily basis of the prevalence of algorithms, since every time that I check on how well one of my books is doing on Amazon or B&N before long I get an email or an internet add suggesting that I buy that book. Then, too, because I live where I can’t just run out and buy a decent shirt, or coat, or even office supplies [since our sole office supply store lost its least ten months ago],and because I have to do that shopping online, I get more “targeted” ads suggesting I buy more of what I just bought.

All of this makes little sense, because I don’t need to buy more copies of the books I wrote. Nor am I likely to buy more shirts after I just purchased some… or more office supplies right after I’ve stocked up.

Now… occasionally I do buy other books, but the recommendations I get from Amazon based on my purchases are laughable. All of this suggests that, while algorithms are being used to extrapolate from my purchases what I might be interested in buying, they’re not doing a very good job… and they’re just irritating.

If that were the only problem with algorithms, I wouldn’t be writing about them.

Algorithms govern the way in which our computers present almost everything to us, from particular ways of seeing the world, reproducing stereotypes, and even strengthening our existing views of the world by tailoring news based on our past reading or searches. In essence, algorithms narrow our view of the world without warning and without providing any sense of what we may be missing.

As ScienceDaily points out, “An algorithm that claims to spot beauty and tell you which selfies to delete implies we should trust technology more than ourselves to make aesthetic choices. Such algorithms also carry assumptions that beauty can be defined as universal and timeless, and can be easily reduced to a particular combination of data.”

Add to that the idea that everything is reducible to data, which in turn affects the way people perceive their environment and everyday relations. This also explains the growing popularity of wearable devices that track aspects of our physical activity and health, then analyze and relay them back to us, directly affecting our behavior.

And last, but certainly not least, there is the fact that there are a host of algorithms that companies and governments use to track the movements and purchases of every cell phone user. A New York Times story in 2012 showed that, using such data, researchers were able to use this data to predict where people would be 24 hours later to within 20 meters.

In 1999, David Brin, both a scientist and an SF writer, predicted the demise of privacy in his book, The Transparent Society. Guess what? We’re there.

6 thoughts on “The Problem with Algorithms”

  1. John Prigent says:

    The results of algorithms can be awkward, too. Some time ago I ordered some underwear my wife wanted that happened to be on Amazon. Bow I keep getting suggestions from there that I might want to buy myself a new bra! It’s just as well that my only computer is a desktop in my house, it would be embarrassing to have those popping up on a laptop in public.

  2. Andreas says:

    Bad algorithms have lead me to start minimising my onlinemail footprint and do as much as possible in person physically.

  3. JM says:

    Someone, somewhere, in some high up management position is currently patting themselves on the back for implementing those online purchase tracking algorithms.

    The rest of us have to contend ourselves with incognito windows and swearing under our breath.

  4. Bob Walters says:

    Anyone who takes that many selfies probably needs help with every day decisions.

  5. Cliff says:

    You should really be using an Ad Blocker on your web browser, for security reason alone. The elimination of adds is just a bonus.

    Should use either Chrome or Firefox as browser, additionally non computer enthusiasts should be using using either a Chromebook or iPad/iPhone devices for internet browsing since they are far safer & more secure than Windows or Mac.

  6. Wayne Kernochan says:

    As an MS in computer science, way back when, I hope you don’t mind my airing one of my recent pet peeves. An algorithm is not (or was not, until recent “marketing” redefinitions), a kind of computer program. An algorithm is an abstraction of a class of computing programs carrying out the same task, in order to identify the “best” algorithm for the task, in general, or in particular cases.

    The classic example is sort algorithms. They all sort numbers. The worst can take inordinate amounts of time to run on non-trivial “numbers of numbers”. The best can perform log n/n times faster than the worst, which makes a big difference in the real world — e.g., in analytics, which uses a sort step as part of running the same operation on trillions of pieces of data efficiently.

    Don’t blame the algorithm for failure to deliver the result you need. Blame the idiotic data scientist or marketer who fails to see things from the end user’s point of view.

Leave a Reply

Your email address will not be published.