Aleem Khan

To know ex­act­ly when, where and for how long a tsuna­mi would af­fect T&T would be worth mil­lions of dol­lars to busi­ness­es, gov­ern­ment and civ­il so­ci­ety. Through com­put­er-gen­er­at­ed sim­u­la­tion, it is pos­si­ble to have a near 100 per cent ac­cu­rate en­act­ment of a tsuna­mi and oth­er nat­ur­al dis­as­ters.

With the help of a Mass­a­chu­setts In­sti­tute of Tech­nol­o­gy (MIT) ocean sci­en­tist, this is what re­turn­ing na­tion­al and Yale-grad­u­at­ed nu­mer­i­cal an­a­lyst Nigel Hen­ry, is at­tempt­ing to sell to the Of­fice of Dis­as­ter Pre­pared­ness & Man­age­ment (ODPM).

"So­lu­tion by Sim­u­la­tion us­es com­put­er mod­el­ing to ba­si­cal­ly probe, and bring in­sight and so­lu­tions to po­lit­i­cal, eco­nom­ic, and so­cial con­texts," Hen­ry said. So­lu­tion by Sim­u­la­tion is the same (and on­ly) com­pa­ny that ac­cu­rate­ly pre­dict­ed the out­come of the To­ba­go House of As­sem­bly elec­tion, pub­lished in the Jan­u­ary 14 edi­tion of the Guardian.

"We use sim­u­la­tions in or­der to solve re­al prob­lems," he said. "The Of­fice of Dis­as­ter Pre­pared­ness Man­age­ment is do­ing a sim­u­la­tion of a hy­po­thet­i­cal tsuna­mi orig­i­nat­ing off the coast of Aru­ba on March 20 (2013). The earth­quake starts at 9 am and the tsuna­mi hits T&T slight­ly af­ter 11 am. I am part­ner­ing with an MIT ocean sci­en­tist (Leg­e­na Hen­ry) and we are giv­ing them what we call a tsuna­mi-in­un­da­tion mod­el. What that does is that it pre­dicts, us­ing com­put­er sim­u­la­tion, ex­act­ly when, where and for how long the tsuna­mi will af­fect each com­mu­ni­ty in Trinidad."

The com­pa­ny will sim­u­late the evac­u­a­tion of per­sons, the move­ment of the trans­port sys­tems, the sea as it comes in and leaves the is­land and the over­loads that the emer­gency re­sponse sys­tems would suf­fer.

"What com­put­er sim­u­la­tion can do is that it can es­sen­tial­ly run a hy­po­thet­i­cal ex­per­i­ment, with­out ac­tu­al­ly do­ing it in re­al life. So with this par­tic­u­lar mod­el, what we're try­ing to do with the ODPM is have an in­ter­play be­tween their sys­tems and their re­ac­tions in terms of mak­ing the var­i­ous calls of pro­to­cols, feed­ing back in­to our com­put­er mod­els, mod­el­ing that over a wider scale and then in­form­ing the hu­man sim­u­la­tion, and go­ing back and forth like that."

Un­for­tu­nate­ly, how­ev­er, Hen­ry said he re­ceived a let­ter from the ODPM say­ing it wants the sim­u­la­tion done but does not want to pay for it.

He said this is an ill­ness with which many com­pa­nies and or­gan­i­sa­tions in T&T are in­fect­ed. They want the in­for­ma­tion but they do not want to pay for it, he said. Asked why he thought this is so, Hen­ry said they may need help mak­ing the link be­tween in­for­ma­tion and the bot­tom line.

The bot­tom line?

"As ex­perts in mod­el­ing the com­plex ef­fects of hu­man be­hav­iour, we could see what is the log­i­cal con­se­quence, or the sim­u­lat­ed con­se­quence of many com­plex in­ter­ac­tions in sit­u­a­tions where you may not be able to do the ex­per­i­ment," he said.

"Let's say the gov­ern­ment want­ed to ini­ti­ate a cer­tain pro­gramme. If it is a com­plex pro­gramme, like one thing leads to an­oth­er that leads to an­oth­er, and you know the rules, you have a cer­tain sense of how the minute ac­tors might re­act, but you wouldn't know what the ef­fect on the glob­al sys­tem would be. We can use a com­put­er sim­u­la­tion to mod­el that, so then you can see what the glob­al or macro out­come is, based on as­sump­tions made us­ing the small in­ter­ac­tions be­tween in­di­vid­ual agents which are usu­al­ly hu­man be­ings," he said.

He said it is sim­i­lar to do­ing mar­ket re­search be­fore launch­ing a prod­uct and spend­ing mil­lions of dol­lars on mar­ket­ing and dis­tri­b­u­tion. To make sure that mon­ey is not wast­ed be­hind some­thing that would not work or would not sell, busi­ness­peo­ple do mar­ket re­search.

"We sim­u­late, us­ing a com­put­er sim­u­la­tion," he said.

The on­ly one of his Ivy-league ed­u­cat­ed sib­lings to re­turn to T&T to con­tribute to the de­vel­op­ment of the coun­try, Hen­ry said he does not pur­port to re­place mar­ket re­searchers, but pro­vides more in-depth analy­sis and mag­ni­tude to their work. He said, for ex­am­ple, the re­sults of a fo­cus group could be tak­en and am­pli­fied to find out, based on the re­spons­es from 10 sam­ple per­sons, what might be the ef­fect if 10,000 peo­ple were in­volved. "From what we know on a small scale, we can use a com­put­er to cal­cu­late what will oc­cur on a wider scale," the 32 year old said.

He be­lieves that if T&T had a cul­ture of sta­tis­tics and nu­mer­i­cal analy­sis like the Unit­ed States, birth­place of his com­pa­ny So­lu­tion by Sim­u­la­tion, it would make a big dif­fer­ence to busi­ness­es. "They would make few­er er­rors," he said.

Sta­tis­tics ben­e­fit every­one

Hen­ry start­ed So­lu­tion by Sim­u­la­tion in 2008, and for most of the com­pa­ny's life­time, served US politi­cians.

At "the US equiv­a­lent of the Cen­tral Sta­tis­ti­cal Of­fice (CSO), the US Cen­sus Bu­reau, and the De­part­ment of Com­merce, all they do is crunch num­bers, and they are very well fund­ed and every month they come out with es­ti­mates of re­tail sales, so the busi­ness­es get to 'free ride' off the gov­ern­ment-fund­ed polls be­cause they are for every­body's ben­e­fit," he said. The US gov­ern­ment un­der­stands it is to the ben­e­fit of all peo­ple if busi­ness­es have ac­cess to re­li­able sta­tis­tics.

"In a sense, I guess the busi­ness­es don't want to put out their mon­ey them­selves. Maybe it ac­tu­al­ly takes state fund­ing in or­der to push it," he said. The cham­bers of com­merce can al­so step up, if the gov­ern­ment is not fund­ing da­ta col­lec­tion, Hen­ry said.

"Busi­ness­es should be able to get to­geth­er and pool re­sources to get these prod­ucts, the sta­tis­tics. The im­por­tant ones, such as re­tail sales by sec­tor, will ben­e­fit all busi­ness­es"

He not­ed that the US puts out re­tail sales fig­ures every month. The US gov­ern­ment al­so pub­lish­es its job­less claims and a host of oth­er num­bers every month.

Hen­ry thinks the rea­son the cham­bers of com­merce may be slow to pro­duce num­bers is that "the start­ing point is so far back that it will be very cost­ly to build the foun­da­tion, to then con­tin­ue." When it comes to sta­tis­tics and nu­mer­i­cal analy­sis, Hen­ry said: "We are re­al­ly far be­hind. They may not see it as fi­nan­cial­ly vi­able," to first up­date the ex­ist­ing sta­tis­tics on the coun­try, and then main­tain that and ex­pand it.

He said that So­lu­tion by Sim­u­la­tion would like to get to the point where it just does what is called de­riv­a­tive mod­el­ing, but it is ac­tu­al­ly hav­ing to do the raw da­ta, and come up with the polls our­selves. If we can't get an in­de­pen­dent poll, we have to fund it our­selves and the cost of run­ning a poll is in the range of five dig­its."

Analysing raw da­ta

Hen­ry said, "if you look at what has worked in oth­er so­ci­eties, the gov­ern­ment will pro­vide most of the raw da­ta, and you will have the busi­ness­es and uni­ver­si­ties do­ing the an­a­lyt­ics on it, which is what could sell."

The pub­lic, he said, whether it is the gov­ern­ment, busi­ness lead­ers or uni­ver­si­ties, have not come to ac­cept the pow­er that nu­mer­i­cal analy­sis can have, and where they ac­cept it in the­o­ry, they don't want to pay for it.

He said the ODPM is "a clas­sic ex­am­ple."

He said to a cer­tain ex­tent he doesn't mind if por­tions of his stud­ies are giv­en away to the pub­lic for free, if he were to ben­e­fit from pub­lic fund­ing, but there are as­pects that can be mon­e­tised and there­fore re­duce the tax­pay­er's bur­den.

Re­turn­ing home to give back

Hen­ry lived, stud­ied and worked in the US for 12 years be­fore he re-mi­grat­ed to T&T about a year and a half ago. He said he came back as "it was al­ways part of the plan to re­turn to help build up the tech­no­log­i­cal re­source base of the coun­try."

He grew up in Port-of-Spain, at­tend­ing Bish­op's Ju­nior School and then St Mary's Col­lege. Both his par­ents are med­ical doc­tors. His fa­ther is a car­di­ol­o­gist and his moth­er is a fam­i­ly doc­tor. His broth­er is an anes­the­si­ol­o­gist and sis­ter, a fam­i­ly doc­tor as well. Hen­ry's par­ents, who al­so re­side in T&T, were al­so Ivy-league ed­u­cat­ed. He is al­so the on­ly one in his fam­i­ly who did not pur­sue med­i­cine.

Hen­ry start­ed off want­i­ng to be a neu­ro­sur­geon. He re­alised he was more in­ter­est­ed in neu­ro­science, so he tran­si­tioned to com­pu­ta­tion­al neu­ro­science, which is build­ing com­put­er mod­els of how the brain works.

He al­so holds a mas­ter's de­gree in in­ter­na­tion­al pol­i­tics from the El­liott School of In­ter­na­tion­al Af­fairs in Wash­ing­ton.

Go­ing for­ward, Hen­ry said his com­pa­ny will place greater em­pha­sis on mi­cro-tar­get­ting which is, in his words, "the use of de­mo­graph­ic and con­sumer da­ta to mod­el (pre­dict out­comes) at the in­di­vid­ual lev­el; who your po­ten­tial cus­tomers will be, af­ter do­ing your tra­di­tion­al mar­ket re­search."

"We will do cut­ting-edge mar­ket­ing tech­nol­o­gy, such as mi­cro tar­get­ting, which in­volves first do­ing the tra­di­tion­al mar­ket re­search to build a cus­tomer pro­file. Once that is done, the mi­cro-tar­geter will build a sta­tis­ti­cal com­put­er mod­el of that pro­file, and be able to look at the pop­u­la­tion at large, with all its mul­ti-faceted spec­i­fi­ca­tions, and then be able to point out those who will be your nat­ur­al mar­ket, and then those who you will be able to ed­u­cate to be­come your cus­tomers, and draw to­ward you, with a lit­tle bit of mar­ket ed­u­ca­tion or per­sua­sion."

?