There’s been a number of reports of issues with the SAM acquisition system at GSA with a recent report that GSA is sending extra resources (some awesome folks) to help fix the issues
It’s easy from afar to complain about problems and we all know how hard it can be to integrate 8 different systems (and generally how hard large scale IT projects can be)
But I thought I’d throw out two ideas to consider for fixing SAM:
1) Try a large-scale challenge – Netflix set up a $1million prize for the first team that could improve their algorithm their predicting algorithm by 10%.
Could GSA set up a $500k prize for the team that integrates the 8 different feeds? Or potentially 2-3 sub contests focused on key problems (one on data integration, one on front-end, one on solving customer service issues, etc)
Too often challenges are used for just fun projects (create a video contest) – let’s release that energy on hard problems like SAM. I’d specifically look at hosting the challenge on kaggle – a huge community of data scientists that does some fascinating work
2) Open Enrollment APIs / Let others bring in registrants – In election season, you can register to vote directly with your county. However there are also lots of new companies like TurboVote that make the process to register to vote super simply and they pass the information on to counties. This allows outside innovation in the voter registration process instead of just relying on the county to be good at voter registration.
What if SAM had open APIs, so they allowed other companies to create registration tools to allow companies to register for CCR or upload certifications (a Turbovote for CCR) – and incentivize those other companies by paying per registrant (GSA pays $5 per company that registers).
IRS successfully did this with e-filing – you don’t have to register/file with IRS directly and instead you file through partner companies that pass on the data (Taxcut, etc). Also during Groupon’s growth that’s how they worked, they told folks they’d pay $3 per new email subscriber and let other companies try unique test/techniques to grow that list and pocket the difference
Got suggestions? How would you approach solving this complex problem?