The 2018 DoDIIS Worldwide Conference 2018 convened in Omaha Nebraska from 12 to 15 August at the CenturyLink Center where the theme was “Data as a Weapon System: Revolutionizing Intelligence.”  There is some metaphysics to unpack when it comes to understanding “data as a weapon system”, which was not deeply explored at DoDIIS 2018, but it seems to me that in the information age data is moving beyond enabling the use of kinetic weapons to impacting combat outcomes where data (including software and algorithms) itself has a material impact on the conflict. All the DoD speakers referenced the National Security Strategy (NSS) in their remarks, noting that data was important to achieving its three tenets of increasing lethality, enabling partnerships, and improving business operations.

My major take away from all the presentations at DoDIIS 2018 is that data is the critical element for mission success by both intelligence producers and consumers (specifically Combat Commands) in the information age.  There was also a strong undercurrent though that the Intelligence Community (IC) is currently not as capable as the U.S.’ adversaries or the domestic private sector at using data effectively.  Several speakers warned that “time is not on our side.”   In his conference keynote DIA Director US Army LTG Robert Ashely observed that what intelligence consumers want and DIA is trying to achieve with its Machine-Assisted Analysis Rapid-Repository System (MARS) is contextualizing intelligence so that it can provide dynamic situational awareness, enhance deeper understanding, and be more predictive.  By “contextualizing” the DIA Director means using data to find other data relevant to the matter at hand and more quickly to the “so what” or what the data is telling the analyst.  Echoing the Director words, DIA CIO Jack Gumtow observed that data is the future and MARS is how DIA intends to “weaponize data.”

In an hour long break out session that was well organized, DIA’s MARS Program Manager Terry Bush and DIA CIO’s Data DISL (Defense Intelligence Senior Level)  Mac Townsend provided a high level overview on why MARS is necessary and what it needs to be able to do.  In an interactive discussion they offered the following:

  1. MIDB was last re-engineered 22 years ago; DIA has been preparing for MARS/MIDB-T for three years
  2. Current Data Governance is inflexible and needs to be changed for MARS to be successful
  3. MIDB touches all parts of DoD as a source of foundational intelligence
  4. The MARS team understands
    1. Today intel producers are also intel consumers and intel consumers are intel producers
    2. Intel producers and consumers are no longer limited to people but include machines as well
  • Intel consumers today operate in multi-domain changing environments
  1. MARS goes beyond MIDB as a data base as it is a data model and architecture with multiple capabilities to make associations MIDB is currently not capable of. MARS must be able to:
    1. Scale to all the data
    2. Must have a flexible data model
  • Be adaptable to provide new capability
  1. And since DODIIS I have learned will need an “object based” storage architecture
  1. Modern caching technology will enable MARS to deliver “data to the edge”
  2. Guard technology will be employed to provide MARS with cross domain security (CDS) capabilities, which I found curious given the migration in the private sector to software defined networks
  3. The MARS team is exploring ledger block technology for access control to data
  4. MARS/MIDB-T represents a “once in an era” opportunity to impact how data for intelligence is ingested, stored, analyzed, and made available and disseminated

From this overview it was clear to me that the DIA MARS Team has a well-developed conceptual understanding of what it wants MARS to be capable of, but is a long way from translating that understanding into clear Request for Proposal (RFP) language that will tell the DIA industrial base what the technical and performance requirements are for MARS and how they will beevaluated.  Based on my experience with NSA’s GREENWAY Program, I expect it will take several RFIs and draft RFPs before a final MARS RFP will be ready.  Given the mission significance of MARs to DIA and the need for near term success, the pros and cons Data as a Service (DaaS) should be rigorously investigated. If successful, MARS will be as important to the broader IC as JWICS has proven to be

In my DoDIIS 2017 summary I observed, that “becoming more data centric vice IT or network centric was an embedded message throughout that conference, but data centric was neither defined nor described.”  This year “data as a weapon system” was not defined and spoken to throughout DoDIIS 2018 as though the concept of “data as weapon system” is well understood.  While all the senior military officers at DoDIIS 2018 carried forward from 2017 the theme that the IC in general and DIA in particular must accelerate its operationalization of Machine Learning (ML) and Artificial Intelligence (AI),  this year I discerned a deeper sense of urgency from them in Omaha.  Sub-rosa there was also a message, particularly but not exclusively from SOCOM’s General Thomas, that alternatives cost effective sources of data are becoming available that have the potential to make IC’ data increasingly redundant.  The IC’s industrial age operating model needs to change to be more like that of information age content providers.

As with DoDIIS 2107, there was no discussion at Omaha this year by any IC senior regarding the NDAA that was signed by President Trump as DoDIIS 2018 was opening, suggesting to me that IC funding is not a leadership concern.   IC ITE was mentioned only by ODNI speakers and without reference to metrics, effects delivered, or costs.  I was surprised that USDI’s pilot Project Maven was barely referenced in terms of how ML and AI can be used to deliver better intelligence effects.   While there was the usual commentary regarding the importance of DIA partnering with industry, no one expressed concern about the work forces at IT technical leaders such as Google, Apple, and a number AI startups not wanting to be associated with furthering national security missions.


  • “Data is the new black” Sue Gordon PDDNI
  • “The IC is not suffering from a shortage of data” Dana Deasy DoD CIO
  • “You guys suck at deep learning and cognition” Google CEO Eric Schmidt as conveyed by SOCOM Commander USA General Thomas
  • “Only two percent of IC data is discoverable” Annette Redmond, State Department INR Director Technology and Innovation
  • “[IC data] management capabilities are not mature enough for the IC to realize ‘data as an asset’” ODNI Study “Data for the IC Enterprise”
  • “MARS is a once in an era opportunity” Terry Bush DIA MARS Program Manager

There were two dominant themes running across all three days of DoDIIS 2018:

Ø  operationalizing process automation, machine learning ML and AI at scale across the IC so that they are inoperative is critical to the IC remaining a value added source of information

Ø  The IC is currently not well positioned either from a data management or technical basis to implement and take advantage of the benefits of ML and AI

I left Omaha confident that the DoDIIS community fully understands that data honed into useful information is the “ammunition” it brings to the fight, but I am less certain that DIA (or any other IC or DoD agency) it is ready to harness all the data available to create the decision advantage at the policy, operational, and tactical level we need to be certain of prevailing in today’s era of 5th generation warfare.  Said differently, MARS truly is a “space shot” for DIA, which is both risky and exciting!”

That’s what I think: what do think?

One comment on “DIA’s “MARS” MISSION

  1. Louis Andre says:

    Joe, as usual, a well-articulated summary of the issues. I’d add (or amplify) a couple of points. As an analyst, I always wanted to “rummage around” in the data — I sought to engage in the “act of discovery,” generate new hypotheses, and optimize the opportunity to apply critical and creative thinking techniques. In the early part of my career, that was possible. In the latter part of my career, it was not remotely possible as the data repositories and streams (the raw material of the analysts’ trade) had been “productized,” for the most part replaced by access to data only after it had been (extensively) filtered, fondled, and packaged. That model must be broken…where possible. A modern approach to data management must feature interoperability at the data level. To achieve that, data must be precisely and granularly categorized and tagged at the content (vice record) level. My two cents.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s