Dan Berkenstock: The world is one big dataset. Now, how to photograph it ...

Prevodilac: Mirjana Sekicki Lektor: Ivana Korom

Five years ago, I was a Ph.D. student

Pre pet godina bio sam doktorant

living two lives.

koji je živeo dva života.

In one, I used NASA supercomputers

U jednom od njih sam koristio NASA superkompjutere

to design next-generation spacecraft,

da bih dizajnirao svemirski brod sledeće generacije,

and in the other I was a data scientist

a u drugom sam se bavio analizom podataka

looking for potential smugglers

i tragao za potencijalnim krijumčarima

of sensitive nuclear technologies.

osetljivih nuklearnih tehnologija.

As a data scientist, I did a lot of analyses,

Baveći se podacima, vršio sam mnoge analize,

mostly of facilities,

većinom postrojenja,

industrial facilities around the world.

industrijskih postrojenja širom sveta.

And I was always looking for a better canvas

I uvek sam tražio širu sliku

to tie these all together.

da sve to povežem.

And one day, I was thinking about how

Jednoga dana, razmišljao sam

all data has a location,

kako svi podaci imaju lokaciju

and I realized that the answer

i shvatio da je odgovor

had been staring me in the face.

bio tačno preda mnom.

Although I was a satellite engineer,

Iako sam bio inženjer satelita,

I hadn't thought about using satellite imagery

nisam razmišljao o korišćenju satelitskih snimaka

in my work.

u svom radu.

Now, like most of us, I'd been online,

Pošto sam, kao i većina, bio povezan na internet

I'd see my house, so I thought,

mogao sam da vidim svoju kuću i pomislio sam,

I'll hop in there and I'll start looking up

popeću se tamo i početi da posmatram

some of these facilities.

neke od tih postrojenja.

And what I found really surprised me.

I ono što sam otkrio me je iznenadilo.

The pictures that I was finding

Slike koje sam pronašao

were years out of date,

su bile zastarele već godinama,

and because of that,

i zbog toga,

it had relatively little relevance

imale su malo značaja

to the work that I was doing today.

za ono što sam danas radio.

But I was intrigued.

No, bio sam zaintrigiran.

I mean, satellite imagery is pretty amazing stuff.

Hoću da kažem, satelitski snimci su neverovatni.

There are millions and millions of sensors

Postoje milioni i milioni senzora

surrounding us today,

koji nas okružuju danas,

but there's still so much we don't know on a daily basis.

ali i toliko toga što ne znamo, svakodnevno.

How much oil is stored in all of China?

Koliko nafte se skladišti u Kini u celini?

How much corn is being produced?

Koliko se kukuruza proizvodi?

How many ships are in all of our world's ports?

Koliko je brodova u svim lukama sveta?

Now, in theory, all of these questions

U teoriji, na sva ova pitanja

could be answered by imagery,

bi se moglo odgovoriti snimcima,

but not if it's old.

ali ne ako su stari.

And if this data was so valuable,

Takođe, ako su ovi podaci toliko značajni,

then how come I couldn't get my hands

kako to da nisam uspeo da nabavim

on more recent pictures?

novije slike?

So the story begins over 50 years ago

Priča počinje pre više od 50 godina

with the launch of the first generation

sa lansiranjem prve generacije

of U.S. government photo reconnaissance satellites.

fotografskih izviđačkih satelita vlade SAD-a.

And today, there's a handful

A danas postoji grupa

of the great, great grandchildren

prapraunučića

of these early Cold War machines

tih mašina sa početka hladnog rata

which are now operated by private companies

kojima danas upravljaju privatne kompanije

and from which the vast majority of satellite imagery

i sa kojih dolazi većina satelitskih snimaka

that you and I see on a daily basis comes.

koje vi i ja svakodnevno viđamo.

During this period, launching things into space,

U toku tog perioda, lansiranje u svemir

just the rocket to get the satellite up there,

samo rakete koja bi postavila satelit gore,

has cost hundreds of millions of dollars each,

koštalo je stotine miliona dolara po komadu,

and that's created tremendous pressure

i to je stvorilo ogroman pritisak

to launch things infrequently

da se lansiranja vrše neredovno

and to make sure that when you do,

i da se, kada do lansiranja dođe,

you cram as much functionality in there as possible.

nagura što je više funkcija moguće.

All of this has only made satellites

To je sve učinilo satelite

bigger and bigger and bigger

sve većim i većim i većim

and more expensive,

i skupljim,

now nearly a billion, with a b, dollars per copy.

sada već oko milijardu dolara po komadu.

Because they are so expensive,

Pošto su toliko skupi,

there aren't very many of them.

nema ih mnogo.

Because there aren't very many of them,

Pošto ih nema mnogo,

the pictures that we see on a daily basis

slike koje svakodnevno vidimo

tend to be old.

često su stare.

I think a lot of people actually understand this anecdotally,

Mislim da dosta ljudi ovo shvata komično,

but in order to visualize just how sparsely

ali da bismo vizuelno predstavili kako je delimično

our planet is collected,

naša planeta snimljena,

some friends and I put together a dataset

prijatelji i ja smo formirali skup podataka

of the 30 million pictures that have been gathered

od 30 miliona slika koje su prikupljene

by these satellites between 2000 and 2010.

sa ovih satelita između 2000. i 2010. godine.

As you can see in blue, huge areas of our world

Kao što vidite u plavom, ogromna prostranstva

are barely seen, less than once a year,

se jedva vide, ređe nego jednom godišnje,

and even the areas that are seen most frequently,

a čak i predeli koji se vide češće,

those in red, are seen at best once a quarter.

oni u crvenom, se vide u najboljem slučaju jednom u tri meseca.

Now as aerospace engineering grad students,

Nama kao studentima astronautike

this chart cried out to us as a challenge.

ovaj grafik je predstavljao izazov.

Why do these things have to be so expensive?

Zašto sve to mora biti toliko skupo?

Does a single satellite really have to cost

Da li jedan satelit zaista mora koštati

the equivalent of three 747 jumbo jets?

koliko i tri 747 džambodžet aviona?

Wasn't there a way to build a smaller,

Zar ne postoji način da se konstruiše manji,

simpler, new satellite design that could enable

jednostavniji, novi dizajn satelita koji bi omogućio

more timely imaging?

podesno slikanje?

I realize that it does sound a little bit crazy

Jasno mi je da zvuči pomalo blesavo

that we were going to go out and just

da smo hteli da prosto

begin designing satellites,

počnemo da dizajniramo satelite,

but fortunately we had help.

ali srećom imali smo pomoć.

In the late 1990s, a couple of professors

Krajem devedesetih, par profesora

proposed a concept for radically reducing the price

je predložilo koncept za radikalno smanjivanje cena

of putting things in space.

postavljanja stvari u svemir.

This was hitchhiking small satellites

To je značilo prenositi male satelite

alongside much larger satellites.

uz one mnogo veće.

This dropped the cost of putting objects up there

To je smanjilo troškove postavljanja stvari u svemir

by over a factor of 100,

za preko sto puta,

and suddenly we could afford to experiment,

i odjednom smo mogli priuštiti sebi da eksperimentišemo,

to take a little bit of risk,

da preuzmemo mali rizik

and to realize a lot of innovation.

a da sprovedemo mnoge inovacije.

And a new generation of engineers and scientists,

I nova generacija inženjera i naučnika,

mostly out of universities,

većinom sa univerziteta,

began launching these very small,

su počeli da lansiraju ove veoma male satelite

breadbox-sized satellites called CubeSats.

veličine kutije za hleb, nazvane Kjubsats.

And these were built with electronics obtained

Bili su izgrađeni od elektronike nabavljene

from RadioShack instead of Lockheed Martin.

u prodavnicama Radio Šek umesto Lokid Martin.

Now it was using the lessons learned from these early missions

Koristeći znanja iz ranijih misija

that my friends and I began a series of sketches

moji prijatelji i ja započeli smo seriju skica

of our own satellite design.

naših sopstvenih dizajna satelita.

And I can't remember a specific day

Ne sećam se jednog određenog dana

where we made a conscious decision

kada smo doneli svesnu odluku

that we were actually going to go out and build these things,

da ćemo zaista početi da proizvodimo te stvari,

but once we got that idea in our minds

ali kada se ta ideja jednom rodila,

of the world as a dataset,

ideja sveta kao skupa podataka,

of being able to capture millions of data points

koji može da obuhvati milione podataka dnevno,

on a daily basis describing the global economy,

koji opisuju svetsku ekonomiju,

of being able to unearth billions of connections

koji mogu da otkriju milijarde veza među sobom,

between them that had never before been found,

koje nikada ranije nisu pronađene,

it just seemed boring

delovalo je prosto dosadno

to go work on anything else.

raditi bilo šta drugo.

And so we moved into a cramped,

I tako smo se preselili u skučenu

windowless office in Palo Alto,

kancelariju bez prozora u Palo Altu,

and began working to take our design

i počeli rad na tome da svoje dizajne izvučemo

from the drawing board into the lab.

iz skica u laboratoriju.

The first major question we had to tackle

Prvo bitno pitanje na koje smo morali da nađemo odgovor

was just how big to build this thing.

bilo je koliko bi veliko sve to trebalo biti.

In space, size drives cost,

U svemiru, veličina proizvodi trošak,

and we had worked with these very small,

a mi smo radili sa veoma malim satelitima,

breadbox-sized satellites in school,

veličine kutije za hleb, u školi,

but as we began to better understand the laws of physics,

no, kako smo bolje razumevali zakone fizike,

we found that the quality of pictures

shvatili smo da je kvalitet slike

those satellites could take was very limited,

koji su ti sateliti mogli da uhvate bio veoma ograničen,

because the laws of physics dictate

jer zakoni fizike određuju

that the best picture you can take through a telescope

da je najbolja slika koju možemo napraviti kroz teleskop

is a function of the diameter of that telescope,

jednaka kvadratu prečnika tog teleskopa,

and these satellites had a very small,

a ti sateliti su imali veoma mali,

very constrained volume.

veoma ograničen obim.

And we found that the best picture we would

I shvatili smo da bi najbolja slika koju bismo mogli snimiti

have been able to get looked something like this.

izgledala ovako nekako.

Although this was the low-cost option,

Iako je ovo bila jeftina opcija,

quite frankly it was just too blurry

iskreno, bila je previše mutna

to see the things that make satellite imagery valuable.

da bi se videla vrednost satelitskog snimka.

So about three or four weeks later,

Tako da smo se tri ili četiri nedelje kasnije

we met a group of engineers randomly

neplanski sastali sa grupom inženjera

who had worked on the first

koji su radili na prvom

private imaging satellite ever developed,

privatnom satelitu za snimanje koji je ikada napravljen,

and they told us that back in the 1970s,

i oni su nam rekli da je sedamdesetih godina,

the U.S. government had found a powerful

vlada SAD-a pronašla snažan

optimal tradeoff --

povoljan kompromis

that in taking pictures at right about one meter resolution,

da praveći slike rezolucije od jednog metra,

being able to see objects one meter in size,

koje omogućavaju da se vide objekti veličine od jednog metra,

they had found that they could not just get very high-quality images,

pronašli su da ne dobijaju samo slike visokog kvaliteta,

but get a lot of them at an affordable price.

već dobijaju mnogo takvih slika po povoljnoj ceni.

From our own computer simulations,

Pomoću naših kompjuterskih simulacija,

we quickly found that one meter really was

ubrzo smo pronašli da je jedan metar ustvari bio

the minimum viable product

najmanji održiv proizvod

to be able to see the drivers of our global economy,

da bismo videli pokretače svetske privrede,

for the first time, being able to count

da bismo, po prvi put, mogli da izbrojimo

the ships and cars and shipping containers and trucks

brodove, i kola, i prenosne kontejnere, i kamione

that move around our world on a daily basis,

koji se svakodnevno kreću po svetu,

while conveniently still not being able to see individuals.

dok, zgodno, još uvek ne možemo da vidimo ljude.

We had found our compromise.

Tako smo pronašli kompromis.

We would have to build something larger

Morali bismo napraviti nešto veće

than the original breadbox,

od početne kutije za hleb,

now more like a mini-fridge,

pre nešto veličine malog frižidera,

but we still wouldn't have to build a pickup truck.

ali ipak nismo morali sagraditi kamionet.

So now we had our constraint.

Sada smo imali svoju granicu.

The laws of physics dictated

Zakoni fizike su diktirali

the absolute minimum-sized telescope that we could build.

konačnu najmanju veličinu teleskopa koju smo mogli sagraditi.

What came next was making the rest of the satellite

Zatim je trebalo sagraditi ostatak satelita

as small and as simple as possible,

u što manjoj i jednostavnijoj formi,

basically a flying telescope with four walls

u suštini, leteći teleskop sa četiri zida

and a set of electronics smaller than a phone book

i eletronikom manjom od imenika

that used less power than a 100 watt lightbulb.

koji bi koristio manje struje nego sijalica od 100 vati.

The big challenge became actually taking

Veliki izazov je bio zaista napraviti

the pictures through that telescope.

slike kroz taj teleskop.

Traditional imaging satellites use a line scanner,

Klasični sateliti za snimanje koriste linearni skener,

similar to a Xerox machine,

sličan mašini za kopiranje,

and as they traverse the Earth, they take pictures,

i dok krstare iznad Zemlje, oni fotografišu,

scanning row by row by row

skenirajući red po red,

to build the complete image.

da bi stvorili potpunu sliku.

Now people use these because they get a lot of light,

Oni se koriste jer primaju dosta svetla,

which means less of the noise you see

što znači manje smetnji nego što se dobija

in a low-cost cell phone image.

u slikama sa jeftinog mobilnog telefona.

The problem with them is they require

Problem sa njima je to što zahtevaju

very sophisticated pointing.

veoma precizno ciljanje.

You have to stay focused on a 50-centimeter target

Moraju biti fokusirani na metu od 50 centimetara

from over 600 miles away

na udaljenosti od oko 1000 kilometara

while moving at more than seven kilometers a second,

krećući se brzinom od preko sedam kilometara u sekundi,

which requires an awesome degree of complexity.

što zahteva neverovatnu složenost.

So instead, we turned to a new generation of video sensors,

Umesto toga, mi smo se okrenuli novoj generaciji video senzora,

originally created for use in night vision goggles.

prvenstveno konstruisanih za naočare za gledanje u mraku.

Instead of taking a single, high quality image,

Umesto hvatanja jedne slike visokog kvaliteta

we could take a videostream

mogli smo uhvatiti video tok

of individually noisier frames,

zasebnih lošijih ramova,

but then we could recombine

no, tako smo mogli ponovo kombinovati

all of those frames together

sve te kadrove

into very high-quality images

u slike veoma visokog kvaliteta

using sophisticated pixel processing techniques

koristeći prefinjene tehnike obrade piksela

here on the ground,

ovde na zemlji,

at a cost of one one hundredth a traditional system.

po ceni jednog stotog dela klasičnog sistema.

And we applied this technique

Ovu tehniku smo primenili

to many of the other systems on the satellite as well,

na mnoge druge sisteme na satelitu,

and day by day, our design evolved

i dan po dan, naš dizajn je prerastao

from CAD to prototypes

od CAD-a preko prototipa

to production units.

do proizvodnih jedinica.

A few short weeks ago,

Pre samo nekoliko nedelja,

we packed up SkySat 1,

spakovali smo SkajSet 1,

put our signatures on it,

potpisali ga,

and waved goodbye for the last time on Earth.

i oprostili se sa njim sa Zemlje.

Today, it's sitting in its final launch configuration

Danas se nalazi na konačnoj lansirnoj konfiguraciji

ready to blast off in a few short weeks.

spreman da poleti za samo nekoliko nedelja.

And soon, we'll turn our attention to launching

Uskoro ćemo se usredsrediti na lansiranje

a constellation of 24 or more of these satellites

sazvežđa od 24 ili više ovakvih satelita

and beginning to build the scalable analytics

i započeti gradnju skalabilnih analiza

that will allow us to unearth the insights

koje će nam omogućiti da obelodanimo uvide

in the petabytes of data we will collect.

u petabajte podataka koje ćemo sakupiti.

So why do all of this? Why build these satellites?

Ali zašto bismo sve to radili? Zašto bismo pravili takve satelite?

Well, it turns out imaging satellites

Pa, ispada da sateliti za snimanje

have a unique ability to provide global transparency,

imaju jedinstvenu sposobnost globalne transparentnosti,

and providing that transparency on a timely basis

a blagovremeno omogućavanje te transparentnosti

is simply an idea whose time has come.

je prosto ideja čije je vreme došlo.

We see ourselves as pioneers of a new frontier,

Smatramo se pionirima novog smera,

and beyond economic data,

koji van privrednih podataka,

unlocking the human story, moment by moment.

otključavaju ljudsku priču, tren po tren.

For a data scientist

Meni kao nekome ko se bavi analizom podataka

that just happened to go to space camp as a kid,

koji je kao mali posećivao svemirsku stanicu,

it just doesn't get much better than that.

nema ničeg boljeg od toga.

Thank you.

Hvala vam.

(Applause)

(Aplauz)