Lying with stats has been around for a long time and the book embedded below is as old as I am..so should prove something:) This book is an easy read and keep in mind some of the language used there was from another time so some of the topics and wording are of how people talked and thought in that era but rather
focus on the methodologies here, a lot of it is still very true; however, it’s gone way beyond this today. I bring this up as we are just swimming in news articles that quote statistics right and left and we can obviously on many determine what’s useful and what’s not and some of what you see is “click bait” to get you to read as all sites are now in a big war of sorts to get your attention. That’s been actually going in for a long time but now with medias looking for new revenue streams, the focus has become intense.
Cathy O’Neil, a former Wall Street Quant, who did modeling work for Larry Summers years ago at DE Shaw is writing a new book called “Weapons of Math Destruction” which is going to be some what of a big update of what it addressed here. She has a lecture series by the same name and you can watch one of those videos in the footer of my blog, video #3. Her new book will be talking about how to “lie with big data”…a little satire but it happens and has been happening for quite a while now with lying in math models
and code. All the algorithms and models run on servers 24/7 and making life impacting decisions about all of us. In the video her comments about Larry Summers are interesting and professional but basically come down to get to the folks that build the models as they are the ones build the tools that economists use to run their spreadsheets, etc. Who better to predict than the ones who create the models? All models are not bad either and there’s some real intelligence to be gained by all means.
These computer scientists build models and it’s good video to hear them speak as these two data scientists in the “Big Data Chat” have backgrounds with mechanical engineering and thus are used to attaching software to tangibles. We can go back to sub prime and I think we can all agree that was the biggest lie with statistics and then some we have ever seen. It could not have happened without models that lied. You can watch video #2 below and listen to some former Wall Street quants tell you how that happened and how the math works. All banks and financial institutions use models so if you want to see how it works with a layman’s explanation, again video #2 will do it for you. What you end up with when models and stats lie are what is called “quantitated justifications” for things that are just not true. I wrote a post a few months ago on that topic along with a couple “must see” videos that will catch you up on what’s happening with news, jounobots and more.
Quantitated Justification For Believing Things That Are Not True And Using Mathematical Processes To Fool Ourselves-The Journalistic Bot Functionality Debuts As Media Can’t Resist the Formulas…
I had to laugh when Cathy, who in her blog post was referencing her mother as she comes from a family of technology and stat professionals referencing this book as she was seeking some information from her mother, who referred her to this book to read as she was too busy “hacking” to go into detail. Think about that, as time moves forward, a lot of the children today will grow up to be able to say my mom or dad is a hacker:) Some can say it now. It’s a good thing we have people that are taking time out to explain 
what “really” happens behind the scenes in a way that the layman can understand as most consumers have not a clue on how this works and it stands to be the biggest attack on consumers ever with technology used to where the other side doesn’t have a thing to bring to the battle ground. I talk about it some here being a former developer from a few notches down with being a former developer as I too have seen queries and other code that presents results that generate money but don’t have accuracy or efficiencies at the heart and I’m not alone, there’s many more out there who can say the same thing.
Until we hold banks and others accountable for what their models do, no much any politician can say or do as the models and code run on the servers and the politicians and government executives are still in the world that verbiage to define laws and rules is still the fix, but it’s not with current day technologies as the verbiage is there but “code runs hog ass wild” as it’s not hard for professional who build models can figure out how to run a model that lies and hides risk and still maintain a legal verbiage status…been going on for years and the code is getting more complex sophisticated. The post below has more details and links about the phenomena that nobody wants to address, which I call Algo Duping and the Attack of the Killer Algorithms. You won’t fix inequality or really even be able to touch it until this part of the equation is addressed and we may have a long time to wait and more dollars for the middle class to lose until this becomes a reality to focus on.
Not Only Is It Time to Hold Banks and Corporations Accountable, But Rather Hold Their CEOs, Math Model Creations and Proprietary Computer Code Accountable As Well…
So if you read the book below, which is not too complex you will be able to recognize how some of this works and how some of what you read today are some of the same techniques and hey
if you want to create a block buster news article, it’s a “how to” lie with your statistics too. What we have ended up with some modern technologies sadly though is the inability to tell the difference between what has virtual value and what has “real value” in the “real world” and all quants know this and again if you watch the videos they explain how “it doesn’t work that way, as people are not like that” (some of the models designed today).
Movie “HER”– Good Example On How Folks Can’t Separate And/Or Unite The Virtual With The Real World, Billionaires Are Made Out of Those Who Know How to Keep These Areas “Gray” With Creating and Maintaining The Addictive Code That Does It…
For those that put a lot of faith in surveys, well watch this Jimmy Kimmel video at this post and watch people in a survey lie their asses off…pretty funny as people do lie as they don’t want to admit that perhaps they are not up to date on certain topics and want to put a good face forward. It lowers the bar all the way down to the “bottom feeder” level if you will…you have to laugh too when you watch it because I’m sure maybe in a smaller way we probably have all done this, self included. Banks and insurance companies have pretty much morphed into big software companies anymore and they model like crazy for profit. If you listen to the CIO of Bank of America, he tells you they are a software and IT company…ready to compete with the likes of Google and Microsoft. Everyone is talking inequality today as you can’t miss it out there but nobody wants to address the technology tools that are used to create it and keep it going. A couple years ago I wrote a series of blog posts called “The Attack of the Killer Algorithms” which addresses this, the rising level of flawed data out
there and how it it bites over and over with using real every day occurrences.
Recommended Reading: “Where’s the Outrage Over Private Snooping?” “The Killer Algorithms Have Teeth & Don’t Care Who They Might Bite
Anyway, I have embedded the book below for a quick and easy read if you want to go there and get in touch with what code running hog ass wild does with little or no regulation and you see the same stuff that was true 59 years ago out there with statistics but much more sophisticated today with the use of modern day models and algorithmic processes. Shoot I keep seeing all these articles on “inequality” and how economics is at fault, well let’s go to the source of the economics and leave the economists in the dirt as they just guess anymore with so much of what you see in some area as all it takes is a couple algorithms to shoot whatever they have predicted right down to the dirt. That’s why restoring Glass-Steagall is just a band aid partial solution as well..it’s more verbiage while code runs hog ass wild as we have “high tech” problems to address that goes beyond verbiage today. In addition you see “small” results from the Financial Consumer Protection agency too as they go after the low tech, low hanging fraud there, don’t want to touch that math. So let’s get into the 20th century and look and see how inequality is actually modeled…there’s folks that create math models and algorithms that put the tools in works for this to occur, they profit and we loose.
Modeling for Inequality With Segmentation, Insurance Industry Uses Backwards Segmentation As Some Models Stand to Threaten Overall Democracy
So now if you read this book, well there are some techniques you might recognize and shoot they still work so jump in the game of “Liar’s Models” if you like (grin). Let’s all lie together for that matter. I hope you enjoy the book as much as I did…good stuff. BD
http://mathbabe.org/2014/02/03/how-to-lie-with-statistics-in-the-age-of-big-data/
0 comments :
Post a Comment