A Turkey on Thanksgiving

Nicolas Chaillan has done it again.  From his new perch outside the Air Force looking in, he has been extremely critical of our military/industrial complex and prescribed an Rx for the cure against China that can win in six months.  And his echo chamber has responded with accolades.  No critical thought has emerged as to the efficacy of his ideas.  He just says them outright.  Everyone is frustrated with the status quo so agile change, under DevOps is the solution.  Of the current comments (137) I read two which may have contained some critical push back.   One, that six months is unrealistic despite the merit of the intent, and two we probably should be more worried about hypersonic systems being developed in China than AI.  I, of course, agree with both of those comments, but there are many more that exist in this article that have been left unsaid in Chaillan’s vision/version of winning with Silicon Valley principles.  I have written my comments here.  

Here is his Od-ed piece if you want to read it.  

The Pentagon needs a new AI strategy to catch up with China | Financial Times (ft.com)

and his Op-ed he posted on LinkedIN

Let's catch-up with China within 6 months (linkedin.com)

I don't want to spend my entire Thanksgiving holiday responding to his Op-ed in the Financial times and his link to this post.  But I do have to make several comments.   Let's start with his academics. Chaillan, of course, cites the problem with China as captured by Christian Brose in the book "The Kill Chain" and implores everyone to read it.  This book ignores centuries of warfighting history and aside from the honor of Brose having worked for, and with, Senator John McCain, falls well short of understanding how we fight wars.  At the end of the book, Brose does capture the political struggles inside the beltway accurately.  It’s worth reading for that reason, but not the warfighting parts. For the record, academically, the other book Chaillan has read, "A Seat at the Table" falls well short of a solution for the DoD. Software agility alone does not win wars even if you couple it with the magic of AI.  Acquisition agility, alone, also does not close the OODA Loop.  If you study John Boyd, in depth, in particular he’s thoughts on both “Maneuver Warfare” and “Destruction and Creation”, rather than his OODA Loop alone, you get a sense of what innovation is really about, particularly when systems must evolve.  And when I say systems, I mean warfighting hardware.  This the hardware in the physical realm that will kill people and break things.  Not the software.

 Also, his recipe for winning can only be applied to the acquisition of things.  And I’ll agree, acquisition is a fun place to be, because what's not cool about building something shiny and new, winding it up, and testing it?  Unfortunately, within our DoD, one could spend their entire career inside a single acquisition channel and never see the system that one has labored on for decades ever be adopted for use.  This is frustrating for those individuals on that evolutionary path, but evolutionarily speaking, entire species spend their entire natural life on a path that ends.  Dinosaurs, for which I am surely one, found that out...until one realizes that every bird around us, came from a dinosaur.  We need many things in development to be agile.  We don’t survive evolutionarily because everyone moves to the mountain in the face of a flood.  We move to the top of the mountain, we move over the mountain, we move to a different shoreline.  Natural selection doesn’t actually choose.  Those who survive are chosen by default.

Organizing, training, and equipping the Department of Defense, alone, does not win wars.  In fact, the entire DoD alone, does not win wars.  National Security is a whole government enterprise.  All instruments of national power must be aligned, along with the will of the people, and brought to bear against our adversaries.  That's not easy to do in any government.  It's very easy to do in North Korea, where you can execute (murder) for example, the students and teachers, who might bring in subversive material (Squid Game as an example), or easier, within a government that can control much of the public dialogue in the press, and mandate more exacting use of standards and protocols so systems can integrate. Even then, if you don't think there is infighting within China (and Russia) about the use of standards for military hardware and the optimization of warfighting effectiveness, you haven't been reading about China (or Russia).

So…if I had the time, I would write an academic paper on each of the things that Chaillan should be checked on…and asked to show us his homework.  John Boyd would ask to see his homework.  My fear is that when we ask, he will tell us his dog ate it.  But here we go…

1) Software is never done- how does this notion not cost more over the lifetime of a system?

2) If you remove IP you remove profit incentives from a capitalist enterprise.  How do hardware companies build hardware for profit that can be taken from them and given to their competitors?  Pretty sure Silicon Valley runs on proprietary IP, the only thing many companies have.

3) DevOps is an unsecure thing.  SecDevOps is putting lipstick on an unsecure thing, what makes SecDevOps secure?  And why can’t we allow our Prime’s do it within their own secure ecosystems?

4) Decision making at the point of work?  That works with widgets, not on the F-35.  How does a software guy working on a fusion algorithm for the F-35 make a decision about the entire aircraft, or more importantly, how a pilot in combat will use that algorithm?

5) Not creating silos fundamentally misses how we fight wars.  Reading books like the "Kill Chain" which doesn't define a single kill chain, will give you this impression.  Explain how netted warfare increases combat power?  This is a fallacy for which many in the NCW business have fallen prey, not just Chaillan and Brose.

6) How can we possibly accept increased risk of insider threat?  First, the insider threat eliminates any chance of #3 above being real.  Second, one insider threat, like Snowden, is all it takes.

7) Just so I’m labeled as a total troll, although I’ll accept the label, one view of the world I agree with Chaillan on is the necessity for increased data sharing...not so much to share widely in both public and private partnership unless under classified agreements/contracts.  But we should find ways to return to the DoD Post 9//11 sharing of data with the IC prior to the Snowden situation (one single insider) and lockdown.  Again, see #3 and #6 above.

I am not denying that China is a threat...  I'm saying that we don't win by catching up to China with AI in the next six months.  It is dangerous if we try.   We win, by continuing to push technology through capitalist principles and solving problems like energy ubiquity.  We need more energy.    Find more energy while continuing to defend and secure our infrastructure while building warfighting hardware optimized for fighting in an effective kill chain.  Then we organize around these kill chains, we already are.  And train relentlessly to fight.  We already do.  To be clear, a kill chain is not a kill web.  Brose is talking about a nonexistent kill web.  To fight everywhere, is akin to defending everywhere.  Must I quote, Alexander the Great?  I will, “To defend everywhere is to defend nowhere.”  The same is true for the offensive projection of force.  That’s an impossible pipe dream…we don’t achieve it with hardware or software.  It’s unachievable.  

In closing, I just have to point out that Chaillan also takes a swipe at DARPA.  That’s truly odd.  Of course, DARPA is not above criticism...however as an organization that defines most of the agility he craves, an organization that has broken the stove pipes, an organization that moves quickly, with the ability to fail often, and with decisions being made at the edge, he totally doesn't understand that organization.  That is a very strange snipe when there are so many other things wrong with DoD Acquisition for him to focus his ire on.