USD(A&S), however, can still disapprove any MTA [Middle Tier Acquisition] program, whether major or non-major. With advisers from all around the Office of the Secretary of Defense, there will be will numerous potential veto points. Each official may extract concessions from MTA programs managed by the services…
Adm. Hyman Rickover’s skepticism to the reforms nearly 50 years ago rings true today. As Rickover wrote to Packard in a memo:
“My experience has been that when a directive such as the one you propose is issued, most of the effort goes into the creation of additional management systems and reports and the preparation of large numbers of documents within the Service to ‘prove’ that the requirements of the directive are being met in order to justify funds for the Service.
That was from my Defense News article, with the editor recommended title “Too many cooks in the DoD: New policy may suppress rapid acquisition.” Read the whole thing.
One of the important points is that USD(A&S)’s absolute veto authority over service programs being designated MTA reverses the trend of delegating decisions down to the services. If the services want to avoid some burdensome regulations, they’ll likely still have to perform a great deal of work keep all stakeholders throughout OSD happy, which are represented on USD(A&S)’s advisory board for MTAs. That may be counterproductive, and perhaps the services will do less MTAs because of it — to keep more authority in-house.
Of course, the biggest issue is finding budget. Given that lining up funds through Congress will take at least 2 years — if everyone already agrees to the plan — acquisitions cannot quite be so “rapid.” All the planning and documentation occuring in those 2+ years will then have the effect of stultifying any application of agile development.
I have to disagree with one of your premises, Eric. In the main article, you write “Programs designated “middle tier” do not have to follow regulatory processes for requirements and milestone reviews. That can shave years off a program schedule.” While this is the frequent mantra of NDIA lobbyists, there is no evidence that it is actually true — especially if you’re talking about OSD oversight processes, as opposed to internal deliberations of the Services.
The sensible view of middle tier acquisition is that it gives permission for the Services to do the easy things without all of the oversight burden that is appropriate to doing very hard things. That is, the point of MTA is not to make any particular acquisition faster, but to make it less painful to do the ones that were going to be fast anyway — and (not incidentally) to incentivize doing fast easy things more often.
The unrealistic view of MTA is that very hard things could be done quickly and successfully if OSD would just let the Services and contractors do what they want without oversight. We have reams of historical evidence that this is false — outcomes on large programs do not improve when oversight is light or absent. If anything, the opposite is true.
If MTA isn’t to make acquisition faster, then the 5 year time limit may be self-defeating — in that it has baked into it backlash from oversight for breach of 5 years. Take Next-Gen OPIR, for example, an ambitious MTA program considering what happened with SBIRS. Then, Congress repeatedly underfunded OPIR compared to requests, almost making sure the 5 year mark will be a challenge.
I tend to think that major systems need to be disaggregated anyway. Making incremental decisions on separate subsystems and then testing out integration, learning, and then doing a major systems approach seems to make the most sense to me. MTA could be used well for each of those steps. Even though the whole system may take well over 5 years, each incremental step in the learning process can take advantage of MTA.
Plus, the smaller the individual effort, the less likely it will receive oversight because it doesn’t have the “prestige” that comes with huge sums and fully defined plans. Oversight is always a good thing, but it may be better done by deciding what programs to cut after some evidence is available (a binary decision) rather than detailed before-the-fact justification, documentation, and management from the higher echelons.
David (or anybody), can you point me to any publicly available research using the historical evidence you mention that conclude outcomes of large programs do not improve when oversight is light? That would be a very interesting read. Thanks!
That’s a great question, Peter — I wish I had a better answer for you. I’m basing my assertion on a lot of past IDA research, much of which is still sponsor-controlled and not publicly releasable. On the technical side, we did many of the Nunn-McCurdy breach root cause analyses for PARCA that were mandated by WSARA starting in 2009. Schedule wasn’t the focus there, but in understanding the causes of cost growth we needed to be able to distinguish management issues from technical issues. IDA also did some studies back in the 80s and 90s on the sources of long schedules and schedule growth, which found that technical problems were the dominant source of both long schedules and schedule growth. The annual program reports from the Director of Operational Test and Evaluation show a similar pattern across many programs.
We also did some work a couple of years ago supporting a CAPE Strategic Portfolio Review, looking at why refurbishing existing systems isn’t faster and cheaper. That work is not public, but again we found that technical issues dominated any effects from acquisition strategy or management or oversight.
In terms of statistical analysis, David McNicol has published a series of papers (all approved for public release) showing that funding climate has a strong impact on cost and schedule growth, and trying to tease out the mechanisms of that. One of those papers looks explicitly at the impact of regulatory regime, but is more focused on unit cost growth and cancellations than on schedule.
Keep in mind that when I say “technical issues”, many of those derive from flawed requirements — either asking for things we don’t yet know how to do, or making assumptions about the initial conditions that turn out to be wrong. In a few rare cases, OSD bears some of the blame (e.g. Joint Tactical Radio System), but in general these are self-inflicted wounds by the Services.
Thank you! I haven’t looked at any of David McNicol’s work, and now that I’m reading the abstracts, I don’t know how I’ve missed him! Thanks!
Eric, appropriately (imo), conflated the concept of a “major” system and an MDAP in the Defense News piece (2nd paragraph under ‘Authority’). This point is way too wonky for the average Defense News reader, but I wanted to highlight for the readers of this blog that according to 5000.80, the threshold for (A&S) having to approve an MTA are the MDAP levels set in Title 10 section 2430; whereas the threshold between major and non-major systems for documentation reasons are set at major system thresholds from Title 10 2302d. Just a little nuance that is lost in Eric’s Defense News commentary. Thanks Eric!
Thanks for the important clarification, Pete. I used the lower threshold bc that is what GAO reported for its stats. Certainly Next Gen OPIR — aka the son of SBIRS — would be an MDAP. But the nuance indeed got lost to me too while writing the piece, so I very much appreciate you putting it right! Luckily I correctly wrote “major weapon system” and “major program” in the right places!