Analytic Modernization

Is There a Sustainable Ops Tempe in 52? Haw Can Deal with the Flaad af Callectian? -- An Interview with (canclusian) FROM: the SlIitaday Editar Run Date: D4i??i2?ll Editar's nate: Here's the canclusian af aur interview with 5113's Deputy Directar far ASLP (pictured). (Ifyaa missed part aae, yaa can find it HERE.) In this segment amments an Analytic Madernisatian and why it is taking center ti. Q: (U) When we last interviewed yau in August 2009 -- yau were Assistant at the time -- yau spake af the importance af creating a sustainable aps tempa in 52. Have yau succeeded? A: (U). Na. Na, I And it's nc-t just because there's sa much happening in the warld. Here's the thing: we have ta take a fundamentally different appraach ta haw we da business. The valume and quality af access we have new gives us an unprecedented cap ability ta praduce same cf the best intelligence we ever have, but the fact is that we are anly getting ta a fractianal partian [if what we have access ta. I can't tell yau far sure that as gaad as what we praduce is, that it's the absalute best we can da. It's nat sustainable far the warkfarce ar the missian ta keep warking in the way we always have. We've embarked an Analytic Madernisatian. In the past there have been it gets tiring! But ta a large degree thase effarts have represented supercharging existing cap abilities -- daing mare with less. We need ta fundamentally change haw we interact with SIG INT. This is a big deal, because far decades we've trained aur warkfarce -- and tank pride in the fact -- that we had the unique resp ansibility within the ta manage this infarmatian in accardance with the 4th Amendment. Ta that extent we thaught af SIGINT as "radiaactive" and it was aurij ta render the infarmatian safe far athers. ISlur pracess was abaut managing the infarmatian in a manner cansistent with the privacy rights af U.S. persans. I believe we have develap ed an awareness af the infarmatian and an understanding af aur autharities that allaws us ta think differently abaut that and in sa daing, ta create an increased capacity and recaver mare time far the I dan't mean anly in but alas in E, in NTCIC, etc. The key is ta better leverage the Intelligence Cammunity and aur partner relatianships far the eaplaitatian af SIGINT, which is samething in the past we wauld nc-t have dane tn the degree we're prapasing -- it was cansidered aur damain [and We must take advantage af the eap ertise and cap abilities in the IC and aur custamer base ta enhance discavery and capacity, and ta make the actic-nable infarmatian available almast as scan as we encaunter it. This is nat abaut turning NSA inta a callectian resaurce far athers [althaugh callectian is in fact ane af aur great and ane we are uniquely qualified ta undertake) -- it's abaut making sure we dan't esp end resaurces daing wc-rk that is net uniquely in nature -- wark that athers can da.

(U) Q: Can yau give an example? A: HEEL) If NEH discnvers a pathway intn an adversary's space and we extract a terabyte nf CAD [camputer-aided design] drawings c-f weapnn designs, is it nf best value far the IE ta have talented NBA wark their way thraugh this 1 terabyte af data? Is it uniquely Or is it af better benefit tn the natian that we enpase it immediately tn the best and brightest weapan designers in the US gavernment ta wark an in a callabarative space, ta triage, assess and explait the value af that infarmatian? In sa daing we are nut precluding NBA fram cantinued access tn the data, any mare than a published SIGINT praduct rep art isn't available far future reference. We are just leveraging the pnwer and esp ertise nf nthers sc- that we can turn nur attentinn tc- that that which only we can explait by virtue af aur unique talents. Q: (U) Isn't it passible that aur custamers wan't want In devote their awn peaple and time In analyzing the data? A: If we get access tn the best infarmatian an a tapic, the value will speak far itself and the custamers will make the resaurces available. It's like YauTube videas -- they "ga viral" when peaple with gaad reputatians recammend a videa ta athers. If we put the infarmatian nut there and manitnr the custnmer resp nnse tn it, we'll knc-w when tn recnmmend specific items they might want tn take a clnse lc-nk at. Amasnn dc-es the same thing by lac-king at hc-w ynu react tn the praducts yau've laaked at. We can figure nut what the cammunities af interest are far that If the material is in a fareign language, we have they can use tn get the gist nf it, and if it lanks prc-mising, they can use the Natinnal 1iv'irtual Translatinn Center tc- get it translated. We want c-ur language facused an uniquely that rarely bail dawn tn a translatinn. St: what dues this appraach accamplish? We've enpased the intelligence ta peaple wha can interpret it and use it, and we've created far cc-llabnratinn. We've alsn nff-lnaded the respc-nsibility tn manage the data in nur respnsitnries and awn the campliance If data is stared in the Library af Natianal Intelligence, sameane else is paying far it. In the end, we esist tn prc-duce The nnly way tn gn mare deeply intn targets is ta avaid getting stuck an praductian that athers can da far themselves. Dur challenge is ta be always nut lacking far samething new. We need ta think abaut prablems, nat just abaut praductian. The callabarative campanent means it's nat an NBA view, a view, Rather, it's npened up tn all an lit-Space. I wnuld like tn see penple lag lit-Space and want tn share traffic and create a multi-seal rep art based an jaint input." Why nat callabarate at the paint af discavery? Q: T. (UHFQUQ) Variaus pushes far analytic maderniaan'an have been going an far decades at NBA, but new the issue really seems ta be taking center stage. In fact, the number ane Gaal far 2011-2015" is to "revolutionize analysis." What's different new? Ft: IKKREL) We live in an Infarmatian Age when we have massive reserves af infarmatian and dan't have the cap ability ta explait it. Iwas tald that there are 2 petabytes af data in the SIGINT System at any given time. Haw much is that? That's equal ta millian 4-drawer filing cabinets. Hc-w many cabinets per analyst is that?? By the end c-f this year, we'll have 1 terabyte c-f data per

secend ceming in. "Ten can't crank that threugh the existing precesses and be effectiye. Q: it's a matter ef yelume? A: Net yelume alene, but alse cemplesity. We need te piece tegether the data. It's imp essible te de that using traditienal metheds. Streng selecters -- like phene numbers -- will beceme a thing ef the past. It used te be that if yeu had a target's number, yeu ceuld fellew it fer mest ef yeur career. Net anymere. My daughter deesn't even make phene calls, and many targets de the same. H150, the cemmercial market demands privacy, and this will drive eur targets te ge maybe inte uneI-tpleitable realms. ISlur natien needs us te leek fer patterns surreunding a particular spet en Earth and make the cennectiens -- whe can de that if net us? And we can't de it using traditienal metheds. B. Q: Leeking inte the future, is there anything that especially werries yeu? eventuality (internal er external) that weuld make it hard fer te centinue te put eut quality intelligence? A: I'm werried that we have se much geed stuff that we ceuld leck dewn and have them just preducing preduct, and semething weuld jump eut and surprise us. Se we need the discipline te invest in the wild and the unkneyms. 9. Q: (U) Is there anything else yeu'd like te say te the SIGINT werkferce? A: (U) At the risk ef being meledramatic, as hard as the werk is and as challenging as the werk enyirenment can be, we sheuld neyer ferget hew lucky we are te have a jeb where we can have se much impact fer the natien. Den't let the turkeys get yeu dewn!