Jennifer Golbeck: The curly fry conundrum: Why social media "likes" say more than you might think

Prevodilac: Ivana Korom Lektor: Mile Živković

If you remember that first decade of the web,

Ako se sećate prve decenije mreže,

it was really a static place.

bilo je to prilično statično mesto.

You could go online, you could look at pages,

Mogli ste da se umrežite, da gledate stranice,

and they were put up either by organizations

njih su postavljale ili organizacije

who had teams to do it

koje su imale timove koji su to radili

or by individuals who were really tech-savvy

ili pojedinci koji su bili prilično tehnološki napredni

for the time.

za to vreme.

And with the rise of social media

Sa usponom društvenih medija

and social networks in the early 2000s,

i društvenih mreža početkom 2000-ih,

the web was completely changed

internet mreža se potpuno transformisala

to a place where now the vast majority of content

u mesto gde većinu sadržaja

we interact with is put up by average users,

sa kojim komuniciramo postavlja prosečan korisnik,

either in YouTube videos or blog posts

bilo preko video snimaka na Jutjubu ili blog unosa

or product reviews or social media postings.

ili recenzija proizvoda ili poruka na društvenim mrežama.

And it's also become a much more interactive place,

Mreža je postala mesto sa mnogo više komunikacije

where people are interacting with others,

gde se ljudi povezuju jedni sa drugima,

they're commenting, they're sharing,

komentarišu i dele,

they're not just reading.

a ne samo čitaju.

So Facebook is not the only place you can do this,

Fejsbuk nije jedino mesto gde je ovo moguće,

but it's the biggest,

ali svakako jeste najveće

and it serves to illustrate the numbers.

i služi kao ilustracija stvarnih cifara.

Facebook has 1.2 billion users per month.

Fejsbuk broji 1,2 milijarde korisnika mesečno.

So half the Earth's Internet population

Dakle, gotovo polovina internet korisnika na planeti

is using Facebook.

koristi Fejsbuk.

They are a site, along with others,

To je internet stranica, koja je, kao i mnoge druge,

that has allowed people to create an online persona

omogućila ljudima da stvore virtualne ličnosti

with very little technical skill,

sa jako malo tehničkih sposobnosti,

and people responded by putting huge amounts

i ljudi su odreagovali postavljajući ogromne količine

of personal data online.

ličnih podataka na mrežu.

So the result is that we have behavioral,

Rezultat toga je da postoje podaci

o ponašanju, izborima, demografiji

preference, demographic data

for hundreds of millions of people,

stotina miliona ljudi,

which is unprecedented in history.

što je neviđeno u istoriji.

And as a computer scientist, what this means is that

Za mene, kao informatičara,

to znači da sam u mogućnosti da stvorim modele

I've been able to build models

that can predict all sorts of hidden attributes

koji mogu prognozirati svakakve vrste skrivenih osobina

for all of you that you don't even know

svih vas, za koje ni ne znate

you're sharing information about.

da ih zapravo delite.

Kao naučnici, mi to koristimo da pomognemo ljudima

As scientists, we use that to help

the way people interact online,

da komuniciraju na mreži,

but there's less altruistic applications,

ali postoje i manje altruistične koristi,

and there's a problem in that users don't really

i problem je da korisnici zapravo

understand these techniques and how they work,

ne razumeju ove metode i kako one funkcionišu,

and even if they did, they don't have a lot of control over it.

a sve i da razumeju, ne mogu mnogo da ih kontrolišu.

So what I want to talk to you about today

Dakle, danas želim da govorim

is some of these things that we're able to do,

o nekim od stvari koje smo mi u mogućnosti da uradimo

and then give us some ideas of how we might go forward

i da dam neke ideje o tome

kako da vratimo kontrolu korisnicima.

to move some control back into the hands of users.

Ovo je kompanija Target.

So this is Target, the company.

I didn't just put that logo

Nisam samo stavila taj logo

on this poor, pregnant woman's belly.

na stomak ove sirote trudnice.

You may have seen this anecdote that was printed

Možda ste pročitali anegdotu u Forbs magazinu

in Forbes magazine where Target

gde je Target poslao flajer

sent a flyer to this 15-year-old girl

jednoj petnaestogodišnjakinji

with advertisements and coupons

sa reklamama i kuponima

for baby bottles and diapers and cribs

za pelene, cucle i krevetiće,

two weeks before she told her parents

dve nedelje pre nego što je ona rekla svojim roditeljima

that she was pregnant.

da je trudna.

Yeah, the dad was really upset.

O da, otac je bio zaista uznemiren.

He said, "How did Target figure out

Rekao je: "Kako je Target znao

that this high school girl was pregnant

da je ova srednjoškolka trudna, pre njenih roditelja?"

before she told her parents?"

It turns out that they have the purchase history

Oni zapravo imaju istoriju kupovine

for hundreds of thousands of customers

za stotine hiljada korisnika

and they compute what they call a pregnancy score,

i izračunavaju ono što nazivaju trudničkim rezultatom,

which is not just whether or not a woman's pregnant,

što nije samo da li je neka žena trudna,

but what her due date is.

nego i kada treba da se porodi.

And they compute that

Oni to izračunaju

not by looking at the obvious things,

ne samo gledajući očigledne stvari,

like, she's buying a crib or baby clothes,

kao što je kupovina krevetića, dečije odeće,

but things like, she bought more vitamins

nego i to da je kupovala vitamine više nego inače,

than she normally had,

or she bought a handbag

ili da je kupila torbu, dovoljno veliku za pelene.

that's big enough to hold diapers.

And by themselves, those purchases don't seem

Same po sebi ove kupovine

like they might reveal a lot,

ne izgledaju kao da mnogo otkrivaju,

ali predstavljaju obrazac ponašanja koji,

but it's a pattern of behavior that,

kada se stavi u kontekst hiljada drugih ljudi

when you take it in the context of thousands of other people,

starts to actually reveal some insights.

počinje da otkriva neke skrivene činjenice.

So that's the kind of thing that we do

To je ono što mi radimo

when we're predicting stuff about you on social media.

kada pokušavamo da predvidimo stvari o vama na društvenim mrežama.

We're looking for little patterns of behavior that,

Tražimo male obrasce u ponašanju koji,

kada ih spazite među milionima ljudi,

when you detect them among millions of people,

lets us find out all kinds of things.

dozvoljavaju da saznamo svakakve stvari.

So in my lab and with colleagues,

Moje kolege i ja smo u laboratoriji

we've developed mechanisms where we can

razradili mehanizme gde smo u mogućnosti

quite accurately predict things

da sasvim tačno predvidimo

like your political preference,

vašu političku opredeljenost,

your personality score, gender, sexual orientation,

rezultat testa ličnosti, rod, seksualnu orijentaciju,

religion, age, intelligence,

versku opredeljenost, godište, nivo inteligencije,

along with things like

kao i koliko imate poverenja u ljude koje poznajete.

how much you trust the people you know

and how strong those relationships are.

i koliko su te veze jake.

Sve ovo mi radimo veoma dobro.

We can do all of this really well.

And again, it doesn't come from what you might

Opet, to ne dolazi iz očiglednih informacija,

think of as obvious information.

kao što bi bilo za očekivati.

So my favorite example is from this study

Moj omiljeni primer je iz ove studije

that was published this year

koja je izdata ove godine

in the Proceedings of the National Academies.

u Zborniku Nacionalnih akademija.

If you Google this, you'll find it.

Ako pretražite internet moći ćete da nađete.

It's four pages, easy to read.

Ima četiri strane, lako je za čitanje.

And they looked at just people's Facebook likes,

Oni su pregledali samo lajkove ljudi na Fejsbuku,

so just the things you like on Facebook,

dakle samo stvari koje ste lajkovali na Fejsbuku,

and used that to predict all these attributes,

i iskoristili su ih da predvide sve ove atribute,

along with some other ones.

zajedno sa nekim drugim.

And in their paper they listed the five likes

U svom izveštaju su nabrojali 5 lajkova

that were most indicative of high intelligence.

koji predstavljaju nagoveštaje visoke inteligencije.

And among those was liking a page

Jedan od lajkova je

for curly fries. (Laughter)

stranica uvijenih prženih krompirića. (Smeh)

Curly fries are delicious,

Uvijeni prženi krompirići su ukusni,

but liking them does not necessarily mean

ali to što vam se dopadaju ne znači nužno

that you're smarter than the average person.

da ste pametniji od prosečne osobe.

So how is it that one of the strongest indicators

Kako je onda jedan od bitnijih indikatora vaše inteligencije

of your intelligence

is liking this page

lajkovanje ove stranice,

when the content is totally irrelevant

kada je sadržaj potpuno nebitan

to the attribute that's being predicted?

u odnosu na atribut koji se predviđa?

And it turns out that we have to look at

Ispostavlja se da moramo da uzmemo u obzir

a whole bunch of underlying theories

gomilu drugih teorija,

to see why we're able to do this.

da bismo saznali kako dolazimo do ovog rezultata.

One of them is a sociological theory called homophily,

Jedna od teorija je sociološka, zvana homofilija,

which basically says people are friends with people like them.

koja kaže da prijatelji imaju zajednička interesovanja.

So if you're smart, you tend to be friends with smart people,

Ako ste pametni, često su to i vaši prijatelji,

and if you're young, you tend to be friends with young people,

ako ste mladi, sprijateljićete se sa drugim mladim osobama

and this is well established

for hundreds of years.

i ovo je već davno ustanovljeno.

We also know a lot

about how information spreads through networks.

Mi takođe znamo mnogo tome

kako se informacije prenose kroz mreže.

It turns out things like viral videos

Ispostavlja se da se popularni video

or Facebook likes or other information

ili lajkovi na Fejsbuku i druge informacije,

spreads in exactly the same way

prenose identično kao i bolesti kroz društvene mreže.

that diseases spread through social networks.

So this is something we've studied for a long time.

Ovo izučavamo već duže vreme.

We have good models of it.

Imamo dobre modele za to.

And so you can put those things together

Kada sve to saberete

and start seeing why things like this happen.

uvidećete zašto se tako nešto uopšte dešava.

So if I were to give you a hypothesis,

Ako bih vam ponudila hipotezu

it would be that a smart guy started this page,

ona bi glasila da je neki bistar momak napravio ovu stranicu,

or maybe one of the first people who liked it

ili da je jedna od prvih osoba koja je lajkovala stranicu

would have scored high on that test.

imala visoke rezultate na testu inteligencije.

And they liked it, and their friends saw it,

and by homophily, we know that he probably had smart friends,

Njihovi prijatelji su to videli

i na osnovu homofilije, pretpostavljamo da je imao pametne prijatelje,

and so it spread to them, and some of them liked it,

pa se sve prenelo na njih, pa su i oni lajkovali,

and they had smart friends,

a i oni su imali pametne prijatelje

and so it spread to them,

and so it propagated through the network

pa se sve takođe prenelo na njih,

to a host of smart people,

pa se sve proširilo kroz mrežu

na veliki broj pametnih ljudi,

so that by the end, the action

i do kraja je postupak

of liking the curly fries page

lajkovanja stranice uvijenih krompirića

is indicative of high intelligence,

postao indikativan za visoku inteligenciju,

not because of the content,

ne zbog svog sadržaja,

but because the actual action of liking

nego zbog toga što čin lajkovanja

reflects back the common attributes

odražava zajedničke osobine

of other people who have done it.

drugih ljudi koji su učinili isto.

So this is pretty complicated stuff, right?

Ovo je prilično komplikovano, zar ne?

It's a hard thing to sit down and explain

Nije lako objasniti prosečnom korisniku,

to an average user, and even if you do,

a čak iako uspete,

what can the average user do about it?

šta prosečan korisnik može da uradi povodom toga?

How do you know that you've liked something

Kako da znate da ste lajkovali nešto

that indicates a trait for you

što odaje neku vašu osobinu,

that's totally irrelevant to the content of what you've liked?

a koja nema nikakve veze sa sadržajem koji ste lajkovali?

There's a lot of power that users don't have

Korisnici nemaju mnogo moći

da kontrolišu kako se ovi podaci koriste.

to control how this data is used.

And I see that as a real problem going forward.

Ja to vidim kao pravi problem.

So I think there's a couple paths

Mislim da postoje dva puta

that we want to look at

koja želimo da razmatramo

if we want to give users some control

ako želimo da korisnicima damo kontrolu

over how this data is used,

nad korišćenjem podataka,

because it's not always going to be used

jer oni neće uvek biti korišćeni

for their benefit.

u njihovu korist.

An example I often give is that,

Jedan primer koji često dajem je da,

if I ever get bored being a professor,

ako mi ikada dosadi da budem profesor,

I'm going to go start a company

osnovaću kompaniju

that predicts all of these attributes

koja predviđa sve ove osobine,

and things like how well you work in teams

npr. koliko dobro radite u timovima,

and if you're a drug user, if you're an alcoholic.

da li koristite droge, da li ste alkoholičar.

We know how to predict all that.

Znamo kako sve to da predvidimo.

And I'm going to sell reports

I prodavaću izveštaje

to H.R. companies and big businesses

HR kompanijama i velikim firmama

that want to hire you.

koje žele da vas zaposle.

We totally can do that now.

To apsolutno možemo sada da uradimo.

I could start that business tomorrow,

Mogla bih sutra da otvorim tu kompaniju,

and you would have absolutely no control

a vi uopšte ne biste imali kontrolu

over me using your data like that.

nad tim kako ja koristim vaše podatke.

That seems to me to be a problem.

Po mom mišljenju, ovo je problem.

So one of the paths we can go down

Jedan od puteva kojim možemo da pođemo

is the policy and law path.

je put pravila i zakona.

And in some respects, I think that that would be most effective,

Donekle, mislim da bi to bilo i najefikasnije,

but the problem is we'd actually have to do it.

ali problem je u tome što bi to trebalo i uradimo.

Observing our political process in action

Gledajući kako naš politički proces funkcioniše,

makes me think it's highly unlikely

izgleda mi malo verovatno

that we're going to get a bunch of representatives

da će gomila političara

to sit down, learn about this,

da sedne i nauči nešto o ovome,

and then enact sweeping changes

a onda sprovede korenite promene

to intellectual property law in the U.S.

u zakonu o intelektualnoj svojini u SAD-u,

so users control their data.

kako bi korisnici kontrolisali svoje podatke.

We could go the policy route,

Mogli bismo da idemo putem politike,

where social media companies say,

gde kompanije društvenih medija kažu:

you know what? You own your data.

"Vi posedujete svoje podatke.

You have total control over how it's used.

Imate potpunu kontrolu nad njihovim korišćenjem."

The problem is that the revenue models

Problem je u tome što se modeli poslovanja

for most social media companies

većine kompanija društvenih medija

rely on sharing or exploiting users' data in some way.

oslanjaju na deljenje i iskorišćavanje podataka korisnika na neki način.

It's sometimes said of Facebook that the users

Za Fejsbuk se nekada kaže

aren't the customer, they're the product.

da korisnici nisu klijenti, nego su proizvod.

And so how do you get a company

I kako da navedete neku kompaniju

to cede control of their main asset

da kontrolu nad svojim glavnim resursom

back to the users?

vrati korisnicima?

It's possible, but I don't think it's something

Moguće je, ali mislim da nije nešto

that we're going to see change quickly.

što će se brzo promeniti.

So I think the other path

Mislim da je drugi put

that we can go down that's going to be more effective

kojim možemo da krenemo, i koji je efektniji,

is one of more science.

je onaj sa više nauke.

It's doing science that allowed us to develop

Primena nauke nam je omogućila

all these mechanisms for computing

da uopšte razvijemo sve ove mehanizme izračunavanja

this personal data in the first place.

ovih ličnih podataka.

And it's actually very similar research

To istraživanje je veoma slično onom

that we'd have to do

koje bismo morali da sprovedemo

if we want to develop mechanisms

ako bismo želeli da razvijemo mehanizme

that can say to a user,

koji bi korisniku rekli:

"Here's the risk of that action you just took."

"Ovo je rizik akcije koju ste upravo sproveli".

By liking that Facebook page,

Lajkovanjem te Fejsbuk stranice

or by sharing this piece of personal information,

ili deljenjem te lične informacije

you've now improved my ability

poboljšali ste moju sposobnost

to predict whether or not you're using drugs

da predvidim da li koristite droge

or whether or not you get along well in the workplace.

ili da li se dobro slažete sa kolegama na poslu.

And that, I think, can affect whether or not

I mislim da će to uticati na odluku ljudi da podele nešto,

people want to share something,

keep it private, or just keep it offline altogether.

da zadrže u privatnosti ili uopšte ne postave na internet.

We can also look at things like

Možemo posmatrati

allowing people to encrypt data that they upload,

i dozvoljavanje ljudima da šifrom zaštite podatke koje postavljaju,

so it's kind of invisible and worthless

da bi bili nevidljivi i beskorisni sajtovima kao što je Fejsbuk

to sites like Facebook

or third party services that access it,

ili servisima trećeg lica koji im pristupaju,

but that select users who the person who posted it

ali da mogu da vide samo odabrani korisnici,

want to see it have access to see it.

za koje osoba koja postavlja sadržaj, želi da vide.

This is all super exciting research

Ovo je veoma uzbudljivo istraživanje,

from an intellectual perspective,

sa intelektualnog stanovišta,

and so scientists are going to be willing to do it.

i naučnici će želeti time da se bave.

So that gives us an advantage over the law side.

To nam daje prednost u odnosu na zakon.

One of the problems that people bring up

Jedan od problema koji ljudi iznose

when I talk about this is, they say,

kada govorim o ovome je,

you know, if people start keeping all this data private,

da ako ljudi počnu da podatke drže u tajnosti,

all those methods that you've been developing

sve metode koje sam ja razvijala

to predict their traits are going to fail.

za predviđanje njihovih osobina će propasti.

And I say, absolutely, and for me, that's success,

Ja se apsolutno slažem, za mene je to uspeh,

because as a scientist,

jer kao naučniku,

my goal is not to infer information about users,

meni nije cilj da nagađam o podacima korisnika,

it's to improve the way people interact online.

nego da poboljšam način na koji komuniciraju na internetu.

And sometimes that involves inferring things about them,

Ponekad to uključuje zaključivanje,

ali ako korisnici ne žele da koristim njihove podatke,

but if users don't want me to use that data,

I think they should have the right to do that.

mislim da bi trebalo da imaju pravo na to.

I want users to be informed and consenting

Želim da korisnici budu informisani i da pristanu

users of the tools that we develop.

da koriste alate koje razvijamo.

And so I think encouraging this kind of science

Mislim da podsticanje ovakve nauke

and supporting researchers

i podržavanje istraživača

who want to cede some of that control back to users

koji žele da deo te kontrole vrate korisnicima

and away from the social media companies

i oduzmu od kompanija društvenih medija,

means that going forward, as these tools evolve

znači napredovanje, kako se ti alati razvijaju

and advance,

i napreduju,

means that we're going to have an educated

znači da ćemo imati obrazovanu i osnaženu bazu korisnika,

and empowered user base,

and I think all of us can agree

i mislim da se slažemo

that that's a pretty ideal way to go forward.

da je to idealan način za napredovanje.

Thank you.

Hvala vam.

(Aplauz)

(Applause)