Re: [xml-dev] Re: XML As Fall Guy

[Kurt Cagle <kurt.cagle@gmail.com>]

On Wed, Nov 27, 2013 at 10:46 PM, Stephen Cameron <steve.cameron.62@gmail.com> wrote:

The only solution is to get people with the expertise to do ALL the work, including writing a specification if necessary, and to take ALL the risk. Financial markets are good at eliminating poor performers, just try getting reinsured after a project failure in the $M category. Risk markets (insurers) get very good at judging what is a reasonable price for something that involves risk. I once worked for a business called 'Objective Risk Management' briefly, that name came from managing risk in mining projects I think.

Diversification is also a risk mitigation strategy, and I think its one that the government (and other users of IT products) may benefit from. Had I been the one to build a health care exchange, I would have sent out an RFP stating the core requirements, then required that one of the deliverables for satisfying the RFP (and to be considered for contention) would be a working prototype that met a core set of requirements. The candidates that were moved to the next stage would receive a $250,000 award, while those that didn't would receive a $25,000 also-ran award. The two or three top candidates would then be expected to implement to a higher set of requirements, and a jury would then award the final contract (worth, say, $20 million) to the winner, who would then be able to bring in subcontractors as needed, so long as that cap was not exceeded. Half of the contract would be paid up front, a quarter at an arbitrary point of completion, and the balance upon successful completion. Failure to complete the contract would mean that the final amounts would never get paid. If the initial candidate cannot deliver to the minimum specifications of the first benchmark, then that contract is ended and the runner-up receives the contract, as well as inheriting the code base from the first. If a vendor comes in under budget and within schedule, they also gain a 10-15% bonus.

This accomplishes several things:

1) It winnows out non-competitive companies while at the same time insuring that small companies can compete against 10,000 person contracting companies (and actually gives a slight edge to the former).

2) It insures that the strongest approaches can be tried and compared, rather than simply who has the best sales team/budget.

3) It makes it clear that there is one company that is managing the project, and all decisions (and responsibility) are made by them, in response to customer requirements.

4) It provides a couple of reality checkpoints, and if a vendor wins the contract but fails to deliver into the midpoint, they lose it, keeping only the money paid to them up to that point. This prevents another common practice - front-loading projects with your best developers, then pulling them once the contract is won (and rates are established) for second tier developers. 

5) It keeps subcontract wage creep down. It's not uncommon on government projects for the same developer to have their rates marked up by 50-100% by their contracting agency, ditto lower level aggregators, and ditto general contractors. This means that a typical developer may be paid $30-50 an hour, but ultimately be charged at $250-$500 an hour, with each aggregator taking their cut, something I've experienced personally (as I suspect others on this list have).

This also gets you away from the preferred provider trap, where you can only compete on projects if you're on a gov't preferred provider list, and the only way you can get on the list is to have completed a project for the gov't (or to have a Senator in your pocket), and it gets you past the lowest bid problem - where the gov't is forced to accept a lowest bid except in cases where the vendor is manifestly unable to complete the project. Instead, you simply ratchet up or down the awards to fit the scope of the project - if you can't get enough qualified vendors, then you probably have to raise the reward. This provides a market pricing structure that's both demonstrable and flexible.