HOW STUPID ALGORITHMS FAILED US > 2008-2013

SEE PART TWO at this LINK

http://jaygaskill.com/BlindFollowersFollowup.htm

HOW STUPID ALGORITHMS FAILED US

LESSONS FROM 2008 TO 2013

ALSO AVAILABLE

http://jaygaskill.com/HowStupidAlgorithmsFailedUsFrom2008Through2013.htm }

Commentary

By Jay B Gaskill

Aircraft, banking systems, political campaigns and medical delivery systems just don’t just arbitrarily crash. There is always a reason. …And it is not always “human error” on the ground level.  Typically, the failure was foreordained by overreliance on “decision crutches” at the management level. This is another name for trusting algorithms over human judgment.

An algorithm is just a complicated recipe, a preset series of sequential steps that, when programmed into an information-processing gadget, enables an otherwise “idiot” device to execute complicated (though sometimes catastrophically inappropriate) “decision-making” formulas without real time human involvement.   Algorithms do not bleed.  But we do.  Real world people always are hurt when algorithms fail.

Yes, there is a need for rapid fire algorithms in emergency situations when human reflexes aren’t quick enough. But we-the-people tend to be lazy (the author is no exception). Because our Brave New World is increasingly complex, we find ourselves relying more and more on algorithms to do our essential thinking for us.

A recurring theme in traditional science fiction was the rogue robot, built without a “kill switch”. We are now relying on algorithms for absolutely critical functions, all too often without any adequate human back up and without an effective human monitoring system.  This is like building autonomous robots without kill switches.

Blind reliance on algorithms is a very foolish and very dangerous mistake. We have already suffered from its consequences.  It is not too late to pull back a bit from the precipice.  But first we to need connect a few dots.

CAN YOU CONNECT THESE DOTS?

  • In 2008, investment bankers blithely relied on complex, impenetrable “asset’ bundles that included toxic loans –concealed/disguised as “assets”. The dirty little secret was that the very complexity of these devices hid an illicit alchemy that seemed to turn fecal matter (hidden bad loans) into gold (but it was fools’ gold).

For an overall review, check out these linked articles-

ü  http://www.usatoday.com/story/money/business/2012/10/02/jpmorgan-lawsuit-bear-stearns-more/1609297/

ü  http://useconomy.about.com/od/glossary/g/mortgage_securi.htm

ü  http://www.cnbc.com/id/36803732 ).

The ensuing asset bubble-burst and the resulting disruptions profoundly damaged the US economy, and almost brought down the entire US banking system. Our country is still struggling to climb out of the resulting economic crater. {And our inadequate recovery is the result of relying on another set of failing economic algorithms, but that is a separate story.}

The takeaway lesson is that overly complex systems can conceal more than they reveal. When we blithely rely on these rigged systems, we are abrogating our obligation to pay attention and make sound choices.

  • The same kind of problem is surfacing in medicine.  Because government oversees insurance, and insurance oversees medicine, all of our trained medical professionals, the physicians and hands-on clinicians are now forced to warp informed clinical judgment in order to conform to bureaucratic categories.  

 

…Algorithms again, in one more form.

Insurance and Medicare “codes” are dumbing down ordinary medical practice.  A number of articles describe the difficulties (sometimes nightmares) of Code Driven Medicine

ü  http://www.nytimes.com/2012/10/09/health/the-ups-and-downs-of-electronic-medical-records-the-digital-doctor.html?pagewanted=all&_r=0 )

“Critics are deeply skeptical that electronic records are ready for prime time. ‘The technology is being pushed, with no good scientific basis,’ said Dr. Scot M. Silverstein, a health I.T. expert at Drexel University who reports on medical records problems on the blog Health Care Renewal. He says testing these systems on patients without their consent “raises ethical questions.”’

ü  http://www.nbcnews.com/health/glitch-medical-code-threatens-mental-health-care-therapists-warn-1B8283085

 

  • The tendency towards “cookbook” medicine (read – “creeping algorithm-driven stupidity”) has reached a crisis point in the mental health field where reliance on a diagnostic “bible”, the so-called “DMS-5” has collided with common sense. Couple this with the insurance code system, and we face an impending disaster.

“Glitch in medical code threatens mental health care, therapists warn.”

ü  http://www.psychologytoday.com/blog/side-effects/201305/the-nimh-withdraws-support-dsm-5

“Just two weeks before DSM-5 is due to appear, the National Institute of Mental Health, the world’s largest funding agency for research into mental health, has indicated that it is withdrawing support for the manual. In a humiliating blow to the American Psychiatric Association, Thomas R. Insel, M.D., Director of the NIMH, made clear the agency would no longer fund research projects that rely exclusively on DSM criteria. Henceforth, the NIMH, which had thrown its weight and funding behind earlier editions of the manual, would be “re-orienting its research away from DSM categories.” …”The weakness” of the manual, he explained in a sharply worded statement, “is its lack of validity.” “Unlike our definitions of ischemic heart disease, lymphoma, or AIDS, the DSM diagnoses are based on a consensus about clusters of clinical symptoms, not any objective laboratory measure.”’

ü  http://www.science20.com/science_20/blog/nimh_delivers_kill_shot_dsm5-111138 }

 

  • Millions of us are now passengers in high performance aircraft that are frequently more controlled by algorithms than by pilots. Machines are making crucial life and death decisions for us, and pilots are undertrained. Is this is dangerous? Ask the passengers of Asiana Flight 214.

In that disaster, an undertrained Korean pilot and his crew blithely relied on an automatic landing system in the San Francisco airport. The plane came in to low and slow, the tail caught on the runway berm, and a fully loaded Boeing 777 was wrecked.

A few days later, an experienced, hands-on pilot, speaking for many senior pilots, sounded a warning in the San Francisco Chronicle.  “Cockpit automation can have a pernicious effect on those skills because the emphasis now in most airline training is on technical understanding – to the near exception of traditional flying and decision-making skills known as ‘airmanship.’

“Now we see a new phenomenon: Experienced pilots, who never fully developed their basic flying skills, passing their airline training with flying colors. These pilots go into the cockpit to perform work that resembles that of a systems monitor more than of the work we associate with an aviator.

“The July 6 crash of Asiana Flight 214 at SFO illustrates the worst potential shortcomings of this new training system which, by the way, motivates airlines to offer early retirement to the more senior pilots in order to replace them with a generation who adores the new technology.”

The author, seasoned Pilot James F. Atkinson, was onto something. I recommend his entire article for close study…because it has across-the-board implications for all of us.

ü  http://www.sfchronicle.com/opinion/openforum/article/Do-pilots-have-adequate-flying-skills-4660853.php  )

POLITICS

Ranked voting formulas (yes, another algorithm) have propelled incompetent, untested leaders into important elected positions with as few as 24% of the votes, all in the name of good government.  Mayor Quan of Oakland, California, for example, won over a better qualified candidate who got almost twice her votes (SEE ARTICLES CITED[i]). As a result, Oakland, CA, which was a beleaguered city, a work-in-regress, is now close to an urban basket case under Mayor Quan highly criticized leadership.

And welcome to the era of checklist voters, who leave their critical judgment behind, and rely on some catechism of Official Party Doctrine (which is yet another algorithm, dumbed down for the harried, hurried voter-bots who want to do the “right thing” but need to be told what that is).

The losing Romney presidential campaign crashed in large part because of the turnout differential.  Mr. Obama’s core constituency of benefitted groups had been conditioned to respond to a perceived political threat to their privileged status, a response pattern that was cultivated over a period of three decades at least.  Democratic turnout could be activated via an existing social network that remained intact between campaigns. Mr. Romney’s campaign geniuses relied on a turnout algorithm (you can’t make this stuff up) in an effort to compensate for the absence of a social network potentially ready to respond to an existential threat to their aspirational status.  The existing social network, the diverse mix of aspiration-motivated people sometimes called the “Tea Party” were unmoved by a weak media campaign.  The Romney turnout algorithm failure has rarely been discussed.

 

THE SOBERING IMPLICATIONS

 

Examples of the improvident reliance of algorithms in politics, medicine, air travel and ordinarily life are far too many to catalogue.  My readers are welcome to send me their own examples to incorporate in the discussion.

All of these examples are about neglecting reasonable judgments (whether about selling bad loans, making intelligent treatment decisions, safely landing an aircraft, or voting for an untested leader) and bypassing actual person-to-person trust relationships in favor of a rote formula, in the name of expediency.

We Americans have come to love such formulas, not only because they seem easier to rely on than our own judgment, but also because they can hide our accountability. “I’m sorry that I had to fire you, but I had no choice,” is easier to say than, “I fired you because you were just not good enough.” …Or, “I’m sorry the voters didn’t turn out after all we did,” is easier to say that “I’m sorry that we did very little street level work to get out the vote.”

The next time you are asked to make an important decision, or to rely on a decision-making or “action” process that consists of “a formula or algorithm”, remember Chesley “Sully” Sullenberger whose skill and judgment as a pilot of US Airways flight 1548 safely brought down a crippled airliner with 155 passengers and crew in the Hudson River in 2009.  No set of computer instructions would have as quickly, intelligently or effectively adapted to that engine failure.

Men and women are not just “in the loop”, we are the loop.  Everything else is a brainless tool, provisionally useful, but not ever to be fully trusted.

Our duty to ourselves, and those loved ones who will follow us, is to push to stay in the game at every level as thinking, informed individual human beings.

This may seem hard, but it is not “rocket science”.

Instead of blindly relying on algorithms, all matters of consequence to our lives must reflect the work of real people, in  real time, in the real world, doing real thinking, making real judgments; and accepting real accountability.

Anything else leads to a series of “train wrecks”. The 2008-13 period was just the beginning….

JBG

Copyright © 2013 by Jay B Gaskill, Attorney at Law

Forwards and links are appreciated. For other permissions and comments, please contact the author by email– outlawyer.gaskill@gmail.com.

 

 

 


Leave a Reply