Big Data in Exec­u­tive Deci­sion-Mak­ing: Friend or Foe?

The role of Big Data in exec­u­tive deci­sion-mak­ing: Data-led deci­sions ver­sus deci­sions made using instinct and expe­ri­ence? Can there be too much evi­dence in an evi­dence-based deci­sion-mak­ing process?

In a con­tin­u­a­tion of her Duet inter­view series, Dr Cal­daro­la, author of Big Data and Law, chats with Prof. Roger Hal­low­ell, an acknowl­edged expert on lead­er­ship, regard­ing the prac­tice of deci­sion-mak­ing today ‑and tomorrow.

Good lead­ers base their deci­sions on hard facts to keep their deci­sion-mak­ing com­pre­hen­sive and objec­tive. A pri­ma­ry goal has often been to have fac­tu­al sup­port for the deci­sion which has been tak­en so as not to be per­ceived as being impul­sive, irra­tional, par­tial or biased. Is Big Data an effec­tive tool for improv­ing this type of deci­sion-mak­ing process?

Prof. Hal­low­ell: In gen­er­al, yes. Good deci­sion-mak­ing should be dri­ven by good data. How­ev­er, a phrase that is fre­quent­ly repeat­ed by the peo­ple in the IT world is: “garbage in, garbage out”. That is why we need to con­sid­er the ques­tion – espe­cial­ly in these ear­ly days of Big Data, AI and the like – What is the qual­i­ty of the data we have? I think his­to­ry has taught us that we need to be scep­ti­cal and hes­i­tant. In this regard, let me give you an exam­ple: In 2003, the US invad­ed Iraq large­ly because we believed that weapons of mass destruc­tion were being hid­den there. Now, why did we believe that? To no small degree, it was because extreme­ly cred­i­ble peo­ple – and I am now refer­ring to Col­in Pow­ell, who was a per­son of the high­est integri­ty – assured us that the best data avail­able indi­cat­ed that there were weapons of mass destruc­tion. Need­less to say, we know now that he was wrong. So even when you have the best data avail­able and you make a deci­sion based on that infor­ma­tion, that data can frankly be wrong. And cer­tain­ly, after the US and NATO expe­ri­ence in Iraq, it is hard to deny it was not a mis­take to have gone in in the first place. And cer­tain­ly, from the view of the Iraqi peo­ple, so many Iraqi peo­ple have died, so many bil­lions of dol­lars have been spent and for what?  Is the sit­u­a­tion any bet­ter today than it was in 2003? No, real­is­ti­cal­ly and hon­est­ly no. In fact, for many peo­ple, it is worse – peo­ple are on the move. So, I think the ques­tion with regard to using Big Data in deci­sion-mak­ing in big organ­i­sa­tions is: To what degree are we con­fi­dent that we are get­ting a defin­i­tive answer? But we must keep in mind that we need to be scep­ti­cal. Of course, it is dif­fi­cult to say that we shouldn’t make deci­sions based on our gut feel­ings but, frankly, we know there is good infor­ma­tion out there which indi­cates that data-dri­ven deci­sions are usu­al­ly and almost always the best way to go. But again “garbage in, garbage out”.  It real­ly depends on the qual­i­ty of the data.

How reli­able can an evi­dence-based deci­sion process be in view of the flood of both fake and true infor­ma­tion – and espe­cial­ly the increas­ing use of agnotology?

For all of your read­ers who do not know what agno­tol­ogy is: Agno­tol­ogy is the study of cul­tur­al­ly con­di­tioned igno­rance or doubt, typ­i­cal­ly to sell a prod­uct or win favour, par­tic­u­lar­ly through the pub­li­ca­tion of inac­cu­rate or mis­lead­ing sci­en­tif­ic data. More gen­er­al­ly, the term also high­lights the con­di­tion where more knowl­edge of a sub­ject leaves one more uncer­tain than before. Active caus­es of cul­tur­al­ly induced igno­rance can include the influ­ence of the media, cor­po­ra­tions, and gov­ern­men­tal agen­cies, through secre­cy and sup­pres­sion of infor­ma­tion, doc­u­ment destruc­tion, and selec­tive mem­o­ry. A clas­sic exam­ple is cli­mate denial, where oil com­pa­nies paid teams of sci­en­tists to down­play the effects of cli­mate change. Pas­sive caus­es include struc­tur­al infor­ma­tion bub­bles, includ­ing those cre­at­ed by seg­re­ga­tion along racial and class lines, which lead to dif­fer­en­tial access to information.

Agno­tol­ogy also focus­es on how and why diverse forms of knowl­edge do not “come to be,” or are ignored or delayed. For exam­ple, knowl­edge about plate tec­ton­ics was cen­sored and delayed for at least a decade because some evi­dence remained clas­si­fied mil­i­tary infor­ma­tion relat­ed to under­sea war­fare. Oth­er exam­ples are the numer­ous sci­en­tif­ic research stud­ies with regard to microwaves, Teflon, plas­tic and the like which man­u­fac­ture doubt about health effects when using these products. 

And here again we receive that infor­ma­tion – whether true or fake, whether well researched or not, whether manip­u­lat­ed or not – from sci­en­tists who are regard­ed as being high­ly cred­i­ble people.

Agno­tol­ogy is extreme­ly top­i­cal because I do not exact­ly remem­ber when it was but, at some point, Pres­i­dent Joe Biden said that Face­book was killing peo­ple for allow­ing so much false infor­ma­tion about the vac­cines against Covid 19. I hate to over­sim­pli­fy, but it real­ly goes back to the fun­da­men­tal prin­ci­ple: Garbage in, garbage out.

My (sec­ond) favourite quote is:

From GIGO to QIQO (Qual­i­ty in Qual­i­ty out)

a term coined by George Fuechsel

If one has false infor­ma­tion enter­ing a sys­tem which is not capa­ble of dif­fer­en­ti­at­ing between true or false infor­ma­tion, then the deci­sion tak­en is only as good as the infor­ma­tion on which the deci­sion had been based.

If one has false infor­ma­tion enter­ing a sys­tem which is not capa­ble of dif­fer­en­ti­at­ing between true or false infor­ma­tion, then the deci­sion tak­en is only as good as the infor­ma­tion on which the deci­sion had been based.

There was a book that came out recent­ly about for­mer Pres­i­dent Don­ald Trump say­ing that he was of the belief that if you repeat­ed some­thing often enough- even if it was false- it became true.

There is also a trend to reduce the com­plex­i­ty of infor­ma­tion by shift­ing the dis­cus­sion to the ter­rain of opin­ions by click­ing on likes and dislikes.

There is also the phe­nom­e­non of gam­ing the sys­tem by fig­ur­ing out how an algo­rithm works, and one comes up with a way of influ­enc­ing those algo­rithms dis­pro­por­tion­al­ly. I remem­ber back in the ear­ly 2000s when the notion of cus­tomer influ­ence about a prod­uct became preva­lent, I start­ed think­ing that this was not objec­tive and was also problematic.

Big Data is cer­tain­ly not a panacea for lead­ers for form­ing their opin­ion and mak­ing their deci­sions. To be sure, US com­pa­nies are mar­ket­ing their com­pa­nies in utter­ly absurd ways by pre­tend­ing they are an “AI leader” and are “pro­vid­ing enter­prise-wide AI”. But what does this mean? Most prob­a­bly noth­ing. Per­haps it sim­ply means that the com­pa­ny in ques­tion is pro­vid­ing AI through a cloud that can be used by oth­ers. But is that real­ly an actu­al val­ue propo­si­tion?  Again, most like­ly not.

It is Impor­tant to con­sid­er the ques­tion of how clear the mes­sage is that data analy­sis is giv­ing us. There is a say­ing by Ronald H. Coase. “If you tor­ture data long enough, it will con­fess to almost any­thing”. Fur­ther­more, Lord Court­ney stat­ed already back in 1895: “There are lies, there are dammed lies and there are sta­tis­tics”. Big Data is not new. The only thing that is new is that data analy­sis is being done more effec­tive­ly than before.

What informs peo­ple when they make deci­sions and form judgements?

Let us begin by con­sid­er­ing the way we tra­di­tion­al­ly got our news in the past and the way so many peo­ple are get­ting their news today. A key dif­fer­ence is that in the past our news was curat­ed by a trust­ed source whether this was the NY Times, the Wall Street Jour­nal, the Finan­cial Times and so on. But there was some­body out there say­ing: This is worth print­ing, and this is not worth print­ing. This is real and this is not real.

The same can be applied to tra­di­tion­al ency­clopae­dias ver­sus today’s Wikipedia. Although Wikipedia does make an effort to ban mis­in­for­ma­tion, the ques­tion remains:  How long does it take Wikipedia or any oth­er dig­i­tal news plat­form to ban mis­in­for­ma­tion? And do they actu­al­ly get around to ban­ning it? Let me give you an exam­ple from Wikipedia: I have a rel­a­tive, who was the Sec­re­tary of Com­merce under Pres­i­dent Eisen­how­er. There is a Wikipedia Page on him, and a rel­a­tive of mine went on Wikipedia and wrote all kinds of things that were unsub­stan­ti­at­ed and absolute­ly wacko. Yet nobody both­ered to cor­rect these entries, and nobody iden­ti­fied this con­tri­bu­tion as being plain wrong and so this page has remained out there in this form.

And what is exac­er­bat­ing this prob­lem? At the moment, when the left wing is being extrem­ist and tries to impose its rad­i­cal views on every­one, the right wing sim­ply shuts down and doesn’t lis­ten.  And, con­verse­ly, when the right wing is being extreme and tries to impose its rad­i­cal views on every­one, the left sim­ply shuts down and doesn’t lis­ten. So, what is cer­tain­ly hap­pen­ing in the US today and in some EU coun­tries as well as, I imag­ine, in some oth­er coun­tries around the world, is that we are only hear­ing from the extreme side and not hear­ing from mod­er­ates, who, quite frankly, are the major­i­ty. This mod­er­ate voice is ren­dered silent – which is a huge prob­lem for society.

Social media plat­forms are earn­ing an enor­mous amount of mon­ey. They are incred­i­bly prof­itable. Shouldn’t they be held account­able for the mis­in­for­ma­tion they are facil­i­tat­ing? Shouldn’t they not use some of their earn­ings to cor­rect this sit­u­a­tion? Face­book is referred to as hav­ing “infi­nite scal­a­bil­i­ty” in its purest form. Once you cre­ate the plat­form you can have as many peo­ple as you want using it. The ques­tion is: Who should be held liable for the dis­tri­b­u­tion of fake con­tent and who should be liable for cre­at­ing fake con­tent. I think that the plat­form dis­trib­ut­ing poten­tial­ly fake con­tent is respon­si­ble for polic­ing itself and curat­ing the con­tent they are dis­trib­ut­ing – they owe that to their cus­tomers. After all, tra­di­tion­al news­pa­pers – like the Finan­cial Times – still car­ry out these impor­tant tasks.

The issue here is that this is a vari­able cost that dra­mat­i­cal­ly reduces the prof­itabil­i­ty of plat­forms like Face­book and that is why such plat­forms are doing every­thing to resist that kind of cost­ly, addi­tion­al work. I think we owe polic­ing to soci­ety; if plat­forms like Face­book, Insta­gram, Snapchat… do not want to police them­selves then I think they should incur the costs for them being mon­i­tored not only with regard to facts/information which they prop­a­gate- but also con­cern­ing the opin­ions they are cre­at­ing. Organ­i­sa­tions that have – at least in part – as their goal to influ­ence peo­ple in a way that mis­guides them should not be allowed to oper­ate. Peri­od. And if these com­pa­nies are in for­eign coun­tries – such as, for exam­ple, Russ­ian sources of mis­in­for­ma­tion – they should be blocked.

The chal­lenge com­pa­nies are fac­ing is that AI – as of today – is not good enough to police fake news or uneth­i­cal behav­iour. There are com­pa­nies out there that have infi­nite busi­ness scal­a­bil­i­ty – mean­ing that they accu­mu­late mil­lions of new cus­tomers by just adding a few more servers for an incred­i­bly incre­men­tal vari­able cost. Nev­er­the­less, cur­rent­ly there is a debate going on con­cern­ing legal­i­ty and ethics when data is processed. Com­pa­nies who do not pledge to be eth­i­cal with per­son­al data will cer­tain­ly starve. The chal­lenge that these com­pa­nies are, there­fore, fac­ing today is the deci­sion whether they are dis­ci­plin­ing them­selves by hir­ing an army of mod­er­a­tors to curate their data and con­se­quent­ly relin­quish prof­itabil­i­ty or whether they wait for bet­ter AI or the breakup of their com­pa­ny (car­tel) into dif­fer­ent pieces, as we know from the his­to­ry of Stan­dard Oil.

But is there a dif­fer­ence between fact and opin­ion? Can opin­ions be banned and curat­ed? And isn’t the inter­net already full of opin­ions in the form of likes and dislikes?

Of course, there is a dif­fer­ence between facts that one can ver­i­fy and opin­ions that are sub­jec­tive. Let us take the exam­ple of the attack on the Capi­tol on Jan­u­ary 6, 2021. Some may be of the opin­ion that it was the right thing to do while oth­ers might state that the inva­sion was ter­ri­ble and vio­lent because three peo­ple were killed. Per­haps we need to make a dis­tinc­tion between opin­ion and fact, as we do in tra­di­tion­al news­pa­pers where we know that we tend to find facts on the front page and opin­ions in the edi­to­r­i­al section.

The same should apply to causal­i­ty. Some peo­ple are con­vinced that peo­ple died due to the COVID-19 vac­ci­na­tion- and indeed there were peo­ple who died short­ly after being vac­ci­nat­ed. Nev­er­the­less, there were no cas­es where the vac­ci­na­tion itself caused death.

I think we as a soci­ety need to edu­cate peo­ple about what fact, opin­ions and causal­i­ty are. It is true that par­tial, incom­plete infor­ma­tion – by cut­ting off an essen­tial ele­ment- can spread and be sold on social media in sec­onds ‑espe­cial­ly if the con­tent is lurid. It is also true that spread­ing mis­in­for­ma­tion is dif­fi­cult to stop because it takes time for research to deter­mine whether a state­ment is com­plete and true.

This is so top­i­cal because there was a quote from a new book about Face­book enti­tled An ugly truth, Inside Face­book’s Bat­tle for Dom­i­na­tion. The book’s name came from a post­ing by a Faceook exec­u­tive who had said in an email that Face­book was all about con­nect­ing peo­ple and some­times those con­nec­tions result­ed in ter­ror­ists’ activ­i­ties, or those con­nec­tions result­ed in peo­ple being killed but the desir­abil­i­ty of con­nect­ing peo­ple was so great that- at Face­book at least- they pre­ferred to err on the side of enhanc­ing con­nec­tion rather than insur­ing truth. This rev­e­la­tion is a smok­ing gun.

What do you think is need­ed in these times of social media, dig­i­tal lead­er­ship and dig­i­tal journalism?

We need to have a new social con­tract in which we say: If you are mak­ing mon­ey as a result of con­nect­ing peo­ple and those peo­ple are look­ing at the infor­ma­tion your plat­form is pro­vid­ing, then that infor­ma­tion must be curat­ed. When you are oper­at­ing in a world where a US Pres­i­dent can speak of “alter­na­tive facts”, we have to admit that this is sim­ply wrong- It is a lie. Dr Antony Fau­ci is the Head of the Nation­al Insti­tute for Aller­gy and Infec­tion Dis­eases, and he is the top White House advi­sor on the pan­dem­ic. Dr Fau­ci was in a hear­ing in the Sen­ate, and a promi­nent Repub­li­can Sen­a­tor said all the evi­dence point­ed to the Coro­na virus hav­ing been released from a lab­o­ra­to­ry in Wuhan. To this Fau­ci actu­al­ly replied: “Sen­a­tor, you are lying; what you are say­ing is not cor­rect.” This is an exam­ple of what we should be doing in gov­er­nance. The major­i­ty of the evi­dence con­cern­ing the ori­gin of the virus point to it hav­ing been trans­mit­ted from an ani­mal to a per­son. There is enough evi­dence to sup­port the idea of the virus hav­ing come from that lab in Wuhan that needs to be inves­ti­gat­ed and I am the first to open­ly admit that. How­ev­er, the real­i­ty is that the pre­pon­der­ance of the evi­dence is that it came from an ani­mal. We need to get peo­ple to under­stand those distinctions.

I like the idea of lead­ers hav­ing to self-reg­u­late them­selves. But the indus­try of social media has proven that they are not doing it. This in turn means that the dam­age they are caus­ing is big­ger than the ben­e­fit they are pro­vid­ing- so you need to reg­u­late that industry.

On the one hand, pri­va­cy is cer­tain­ly not meant to cost any­thing yet, on the oth­er hand, it is the nature and indeed the duty of busi­ness lead­ers to earn mon­ey. Mil­ton Fried­man said once: “The only objec­tive that busi­ness lead­ers have is to take care of their share­hold­ers”. Nev­er­the­less. there is also the oth­er side of the spec­trum. The US Cham­ber of Indus­try and Com­merce has recent­ly stat­ed that com­pa­nies should take care of a vari­ety of share­hold­ers ‑includ­ing cus­tomers and employ­ees. Gold­man Sachs just report­ed that com­pa­nies not hav­ing a wide enough view of their stake­hold­ers will be divest­ed. That is the new lead­er­ship challenge.

So, allow me to reit­er­ate at this point that Big Data is not a panacea and it is not even new. That hav­ing been said, we can cer­tain­ly enhance its use as well as our abil­i­ty to analyse it by employ­ing pow­er­ful sup­ple­men­tary forms of tech­nol­o­gy, such as AI.  The qual­i­ty of data mat­ters and the result must be a clear mes­sage. Final­ly, lead­ers should curate their data and infor­ma­tion (with or with­out AI) and need to be engaged in a con­ver­sa­tion about Big Data and pri­va­cy with their constituency.

Prof. Hal­low­ell, thank you for shar­ing your insights on the prac­tice of deci­sion-mak­ing today and tomorrow.

Thank you, Dr Cal­daro­la, and I look for­ward to read­ing your upcom­ing inter­views with recog­nised experts, delv­ing even deep­er into this fas­ci­nat­ing topic.