Open Source Micro-purchasing: An experiment in federal acquisition
18f.gsa.gov
18f.gsa.gov
Are you expecting most companies applying to this to actually be small shops?
Are the scale of projects going to generally be designed/selected to take much less than the allotted 10 business days?
> For the first iteration of this experiment, we’ve made a decision to limit eligibility to companies that are registered on SAM.gov at the time of bid.
I think for this experiment to work in the long run it needs to be handled in a way that makes it clear:
A) You don't need to be a registered LLC/Corp/etc.
B) You don't need to invest in any kind of liability insurance for your contribution.
C) It can be staged in a way that it could be ~60 hours/month with multiple "purchases" for longer projects.
It'll be interesting to see if this works. I rarely found it worth my while engaging in a job with a new client if it billed under 10-15 days. $3500 is only a few days work, so they'll have to ensure the administrative overhead in bidding and billing is minimised.
> Whoever has the lowest bid at the closing time will have 10 working days to ship the code necessary to satisfy the criteria. If the criteria are met, the vendor gets paid. It's that easy. If the criteria aren't met, the next lowest bidder gets 10 working days to ship the code. In order to make this work, we will provide acceptance criteria, instead of requirements. This limits the possibility of misinterpretation and ensures the quality of delivery.
I fear that the focus on tight acceptance criteria may encourage gaming the system. There's no incentive for the contractor to do more than the bare minimum. Even if you gain a reputation with the procuring team of being reliable they still have to go with the lowest bidder for the next job.
For contractors, the optimum strategy would seem to be bid without spending any time coming up with an implementation strategy, then drop any jobs you win that don't seem worthwhile after a few hours working out how to implement it.
Ultimately this becomes a test of how well the jobs can be specified. For this to be efficient they have to include the time spent internally preparing and assessing the result vs working over a longer periods with people with whom they can build a trusting working relationship.
It might be more efficient to rely less on price competition and more on reputation. That would make it a low overhead system for engaging trusted contractors on individual tickets. There'd be far less preparation required in defining very tight acceptance criteria.
Assigning reputation in a system with financial incentives is a challenge in itself too. If reputation is decided in a non-transparent process, such as members of 18F deciding, the system is ripe for bias, e.g. towards companies with ex-collegues, friends or aquaintances. Any public form of online voting is subject to manipulation, closed voting of the companies involved would leat to a conflict of interest as companies have no incentive to distribute reputation fairly, and raising and lowering reputation depending on if the job was done or not allows companies to do the contract with the lowest effort possible, leading to the problem the above poster described.
Watch this space, with enough marketing, you are about to see all the 1000s of companies constructed with no real skill, only certifications (Veteran-owned, HUBZone, Woman-owned, Alaskan, Minority, etc), flood the market with unrealistic bids.
The critical path issue with government procurement of information technology is not cost alone... it's with the insistence of the customer to require a custom solution, and the inability to describe the requirement, and thus, technical acceptability.
What this project will do, is show that even when boiled down to very micro, unscalable size, "lowest price, technically acceptable" (LPTA) is an inappropriate purchasing pattern for non-commodity information technology, or more any non-commodity, really.
Micro-purchases exist to allow agencies/offices to make purchases of more common things without going through normal acquisitions procedures. This might be a software license or office supplies or a new projector for the conference room. Here, they're using it for software development time. It's a faster way to spend money than when purchasing anything over $3500, which means you (usually) get the product/service faster as well.