Some News Leaves People Knowing Less

TtWYF.jpg 496×518.

For the next time someone asks.

Awesome and terrifying at the same time. The PS3 possibilities alone are scary.  Also, they forgot the example from started Stargate Universe where Eli solves the algorithm that Dr. Rush was embedded in the game.

http://www.computer.org/cms/Computer.org/ComputingNow/homepage/2011/0711/W_IT_EnderWiggin.pdf

Interesting read.  Lots of misunderstanding all around – both ways.  $2.7B over 4 years.  Probably wrong $ as well – it’s been funded at least since ’99 as it’s own program.  Looks like they are applying the entire funding of all pieces (at least 10 distinct large pieces) to the single analytical portion.  The 2011-14 costs estimated at ~$794m for the entire program. (https://acc.dau.mil/adl/en-US/436075/file/56596/DCGS-A_MAR_20101231.pdf)

globalsecurity: http://www.globalsecurity.org/intell/systems/dcgs.htm

Army view: http://defensesystems.com/articles/2011/03/29/project-manager-view-army-distributed-intelligence.aspx

GD Operational view: http://www.gdc4s.com/content/detail.cfm?item=004279f8-a9a9-4c18-8051-f9e1acb643ee

If anyone thinks the data coming in from a warzone is in anyway clean, metatagged, geotagged properly, accurate, standard, single spoken language, single type (text, audio video, signals), or timely is kidding themselves.  Hell, network coms aren’t even there all the time so you have online and offline versions of a lot of stuff.  Can you imagine using any current commercial tools when they have satellite lag time, intermittent coms, <64k bandwidth, hundreds of disparate data feeds, 4 year old computers, must be highly encrypted, secured, filtered for need to know, deployed and maintained in a warzone, audited for everything, and done with a budget (including, deployment, maintenance, development, testing, security lockdowns, massive amounts of equipment, coms, and more) for less in 5 years than Google spends on R&D in a single year?

Then for analysts to do their job using the tool with all this data tied together, interpreted, and linked, they have to find the needle in the haystack, the one thing that didn’t show up in the keyword search with 10,000 other results – and do it in near real time.  Not any easy job to do with any tool.

Not to say that DCGS doesn’t have big issues with performance, functional focus, ease of use, and scope.  A good chunk of it is old and outdated (’99 remember?)  Probably a lesson in needing rewrite and conceptual focus.

A note on scope being huge – for example Google Earth may be great at tracking and showing where things are/were, but that would really be just one of hundreds of requirements of DCGS.  Scope is the bane of Govt. systems, they often grow out of older software that gets added onto for decades with multiple contrators and mashed together with multiple other systems as well.

The ad for Palantir at the end is no better either (diminishes the article) – another in-q-tel funded system pushed on the IC that meets part of the analytical requirements.  All of these tools do better after the fact then in real time anyway (which is highly valuable and Palantir is probably good at it given the right data feeds and good analysts).  But, predictive/preventative analysis has a long way to go and these tools don’t do it well yet - modeling is hard.  After the fact is nice for stopping fraud or other ongoing activities from continuing, but really doesn’t necessarily prevent the next upcoming terrorist attack where indicators aren’t identified ahead of time (easy after the fact).  Has anyone asked how much the DoD and IC already spent on Palantir? Look at their VC and revenue.  Maybe it’s a good thing, maybe not.  Duplicative work isn’t necessarily a bad thing.  Lot’s of innovation comes that way.

http://kellblog.com/category/palantir/ is an interesting read.

Enough…