For the past few weeks, GSA's tech team 18F has been running a "reverse-auction" for code.
The General Services Administration is struggling to pinpoint how much the federal government should pay for open-source software code.
But this time, at least, it's not that it's too expensive. In fact, as evidenced in its new micropurchasing experiment, open-source software coding services could be too cheap.
For the past few weeks, GSA's tech team 18F has been running a "reverse-auction" in which developers state the price for which they'll write bits of open-source code for federal customers. Customers choose the lowest bidder, who then has 10 days to complete the task.
If the product is inadequate, agencies move on to the next lowest bidder. (18F is experimenting with the "micropurchasing" system, which allows agencies to make purchases for products or services less than $3,500 directly from the vendor.)
In its first trial, 18F encountered an unexpected bump when a vendor -- Brendan Sudol -- bid just $1 to write open-source code, according to the 18F blog. The 18F team members initially suspected the $1 bid was a mistake -- they awarded it once they confirmed its validity -- but worry that low bids like this could force them to rethink the reverse-auction model.
"If it's always going to be $1, then people aren't going to start coming to bid," staffer V. David Zvenyach told Nextgov. "We want this to be sustainable . . . The question is, 'How can we structure these micropurchases going forward to make sure that people are willing to bid, and that the bids are at a sustainable rate?'"
Though this experiment was limited to companies already registered through SAM.gov, the federal "System for Award Management," Zvenyach added that 18F hopes to tap a broad swath of companies specializing in open-source code, some of whom develop that code for free, which could drive down the price for those services.
The bidding system itself presented its own problems for 18F, Zvenyach wrote in the blog post. When it first announced the "micropurchasing" experiment in mid-October, 18F had planned to release an online platform for the bids by Oct. 26, but it wasn't ready by then. Given a time shortage, 18F used a Google Form to collect bids. (Zvenyach wrote the Google Form exhibited a glitch that could have prevented some bidders from knowing if they were the leading offer.)
"This is the reverse of how it should have been: We should have started with the Google Form and then worked on the platform," he wrote. "We optimized too early. But we’re acknowledging it here, and we’ll learn from it on the next iteration."
The micropurchasing experiment is part of 18F's broader effort to get small and nontraditional contractors to sell to the federal government. In August, for example, 18F unveiled its Agile Blanket Purchase Agreement, pre-approving 16 firms to sell "agile development" services to government agencies.
Despite some obstacles, Zvenyach wrote, the micropurchasing trial has "validated the core concept that open-source micropurchasing can work, and it’s a thing we should try to do again."
In October when 18F first announced the experiment, Zvenyach had warned readers "this might be a terrible idea."
"But at 18F, we're committed to experimentation, and we want to see what happens," he wrote on 18F's blog at the time. "Our hypothesis is that vendors can ship great code under the micropurchase threshold, and we see opportunity to use procurement authority in new and productive ways. . . If it works, that would be fantastic. If it doesn’t, it'll be an inexpensive experiment and we will have learned some new things."
Zvenyach told Nextgov the next iteration of the experiment would likely be announced in the next couple of weeks.