DARPA and the service research labs achieve amazing results. Have you heard about electronics that will physically vanish on command to prevent adversaries from copying our designs? There are many astounding experimental projects. So why do programs seem to fall flat when they try to develop those experiments into fully operable and producible systems?
The DOD itself knows the root of the problem:
The PM community often has a difficult time synchronizing the technology transition funding. The PPBS requires a two-year lead-time for funding to be proved. As a result, accommodating fast-changing S&T developments in acquisition programs can be a challenge. The PM community cannot always predict the pace of innovation two years in advance, and funding may not be available for fast-moving S&T projects that are ready for transition. Therefore, a desirable S&T project may stall for 18 to 24 months, awaiting funding. This gap is sometimes called the “valley of death.” [Emphasis added.]
This is not the same “valley of death” that you hear about in the commercial world.
In the commercial world, the “valley of death” refers to the time between product development and consumer marketing.
Not so in the DOD. Production and fielding of an entire inventory is decided upon before the darn thing has ever been developed. It isn’t like the commercial world because:
(1) there is only one buyer (monopsony), though it is made up of dozens or hundreds of offices, each of which with some power to veto a project
(2) the buyer decides product requirements before development occurs (more likely deciding technical specifications)
(3) the buyer fully funds product development at the buyer’s own risk
(4) production and operations are committed to before system development starts, and regardless of test outcomes
(5) the seller’s business operations are in some cases decided by the buyer
Marketing a system to enter the force structure occurs before an operable system is ever created. Generally they are bets on future technology given some experimental indications from the labs.
When the GAO reported on why so few technologies make it across the “valley of death,” it basically answered (like always) that there needed to be more processes:
DARPA should regularly assess technology transition strategies, refine training requirements, and increase dissemination of technical data for completed programs. DOD did not agree to take GAO’s recommended actions, which remain warranted, as discussed in the report.
No. It is what the DOD reported on above. The budget process plays a focal role.
The program budget in the DOD controls plans and activities, and not the organizations or objects of expenditure. It is the instrument of central planning.
The program budget idea became merged with Armen Alchian’s ideas sequential decisionmaking. The outcome was never what Alchian intended.
The RDT&E appropriation in the program budget has Budget Activities (BAs) that determine which stage of development a project is in. BAs 6.1-6.3 are traditionally the S&T areas. BA 6.4 is where prototyping starts after Milestone A. Then full-scale development of a producible/operable system starts in BA 6.5.
DARPA and the technology labs are only funded through BA 6.1-6.3. The hand-off often occurs at usually BA 6.4, sometimes BA 6.5, where specially created program offices handle all further aspects of development, test, and production.
 |
| 2017 enacted RDT&E budget by Budget Activities. Note that the S&T activities 6.1-6.3 are only 18% of the entire RDT&E budget. |
Now, projects even in BA 6.1-6.3 are given line item attention in the budget. But they don’t become tightly controlled as a Program of Record until generally after the Milestone A or B hand-off to BA 6.4-6.7.
Programs of Record require a consensus from 50 or more offices for funding approval. But this is exactly contrary to the rivalrous exploration that occurs in markets.
Marc Andreessen has stated that non-consensual ideas are the only ones that promise revolutionary change and big returns. George Schairer, former VP of Boeing, said something similar in 1975:
Anything that the majority agrees to probably is wrong for tomorrow. It is right for today, but probably not right for tomorrow. I wonder about such wild ideas as you would ever fly an airplane with a jet engine or have an atomic bomb or radar, or many of the great things we base our defense upon. At the time they were initiated, certainly any group of 10 people you could have get 10 together, presumably knowledgeable, would probably have voted them all down.
 |
| Offices involved in major Air Force program decisions. Some offices are listed once, but have multiple divisions that have separate inquiries. This is just within the DOD, and excludes non-DOD decision makers. |
The worst part of this story is the misalignment of the acquisition and budgeting processes. The acquisition process proceeds from Milestone A (tech development, like prototypes), to Milestone B (full-scale development), to Milestone C (production). These milestone decisions are supposed to authorize work into the next sequential stage of development.
However, for funding to be there, the milestone decision must be made two years ahead of time through the PPBS. You need to have program information and approval two years ahead of time to allow one year for budgeting procedures. Then you have the funds authorized, which are very difficult to reprogram should some risks or opportunities materialize.
The PPBS severely restricts flexibility. It requires explicitly detailed plans and promises justified to many offices. It excludes considerations of trust and responsibility. For a fantastic discussion, see Chuck Spinney’s statement to Congress.
This is why only big defense incumbents survive. Only they can carry high levels of overhead for years on end until a major development gets approved. Then they can recoup their IRAD and B&P costs through overhead because production and operations contracts are basically assured.
 |
| The milestone acquisition process, outlined above, authorizes programs into new phases of development. The “valley of death” is generally at Milestone A. Milestone authorization does not authorize funding through the budget. The acquisition cycle is event based. The budget cycle is calendar based. New programs need 2 years lead time to get a line item in the budget, so long as all bureaucratic approvals go smoothly. |
The DOD has a highly centralized decision structure. You can decentralize milestone decisions to the lower levels, as has been done from time to time. But the program budget is the only place where funding will be made available, and that gets everyone involved from the service staffs to headquarters, OSD, the President, Congress, the GAO, and so forth. No wonder ideas die before they become a Program of Record. You need a bunch of outsiders to say whether a undeveloped technology is worth it or not.
To solve the “valley of death” problem, strike the program budget. An organizational budget provides flexibility and personal accountability. Production decisions cannot be merely a done deal with development approval. The “valley of death,” if anywhere, should be between fully tested systems and production, not between S&T experiment and full-scale development.
This is just scratching the surface of the technology transition problem, but I think that the emphasis on the budget process is fundamentally correct. Problems in acquisition or requirement are merely symptoms of the program budget.
Why should we care about the effect of the PPBS on technology development? The inability for the DOD to be adaptive to uncertain technologies could be one of the biggest drivers of the cost disease problem in defense. It relates to the sustaining vs. disruptive technologies framework.
Leave a Reply