The 2008 Financial Crash posed a stark reminder about how out-gunned government regulators were compared to the big financial services companies that leveraged state-of-the-art computer technologies to design super-complex deals and make lightning-fast trades. It was like one used machine guns while the other fought back with bows and arrows.
Tim O’Reilly has some big ideas about how to rectify the situation by dramatically modernizing the entire notion of government regulation. He talks about “algorithmic regulation” that harnesses computer power much like top tech companies in Silicon Valley do. Those overseeing Google search, for example, set their sights on ultimate outcomes (getting accurate answers), try multiple paths forward, constantly runs tests and study the data, and incrementally make adjustments to fine-tune the best results. Government regulators could use a similar iterative, data-driven approach.
After all, regulation is simply a system for keeping things working right, keeping on track towards a desired outcome. We can reinvent 21st-century versions of regulation that work much better and take advantage of all our current tools. O’Reilly has another concept of peer-to-peer self-regulation, where citizens increasingly take over from paid government bureaucrats, like relying more on citizens making Yelp-like reviews to keep an eye on restaurants.
O’Reilly admits he does not have all the answers, but good starting points for a very important topic. We expect to draw together a terrific roundtable of innovators from many perspectives to build on these exciting new ideas. Full Post →
Tim O’Reilly, who we once described as a model reinventor, lived up to his billing in what turned out to be a model roundtable too. Tim hit the highest bar that we strive to achieve in this roundtable series. He came to the table with some big but unfinished ideas about how to reinvent government regulation by borrowing new approaches and best practices from the Silicon Valley technology world. He handpicked the exact people he wanted at the table to help him (and us) learn how to improve on those nascent ideas. Our lively 90-minute conversation did iterate the ideas and together we moved the ball forward, ending with some concrete suggestions for next steps.
It all started with Tim making the case that quite a few technology companies in Silicon Valley are essentially in the business of regulation in the broadest sense. Google delivers accurate search results by regulating the process and ensuring that bad guys who try to game the system with spam lose out, while those websites that play by the rules succeed. The way they regulate largely differs from how government regulators work. Google is constantly trying to improve its regulation by running tests, collecting and studying vast amounts of data, and then incrementally adjusting the rules. O’Reilly thinks governments could try much more of this iterative, data-driven “algorithmic regulation.”
Many tech companies also use reputation systems to provide a regulation of sorts. If customers give poor ratings to a room to be rented on Airbnb, then there’s no need for a government inspector to pronounce it a fleabag. O’Reilly proposed that we try to figure out where this new kind of participatory regulation might replace the more bureaucratic one.
We quickly moved to a nuanced conversation that went into more detail about how the tech companies actually carry out these strategies. This was driven by two top tech experts: Matt Cutts, a distinguished engineer at Google, heading the webspam team, and Jason Tan, the CEO of Sift Science, which helps companies fight fraud through big data analytics. We also had a top innovator from inside government explain how open the system did appear for change. This came from Sophie Raseman, the Director of Smart Disclosure, Office of Consumer Policy, from the US Treasury Department.
The key is to extrapolate principles that could be applied more broadly to various government regulatory situations:
- Tim gave some, starting with focus on outcomes at all times. That should be the guiding star that structures everything else. If the original rule does not reach the desired outcome, then adjust that rule to reach that outcome.
- Another one: let data guide the regulation. Study the data to understand what works and then adjust the regulation in light of that data.
- Both those imply that regulation should be more iterative. Most current regulation gets set as law or policy, and then stays static and brittle until everything gets re-evaluated years later.
- Sophie added the principle of openness. Try to open as much government data as possible and let others outside government help find ways to make it accessible (machine readable) and useful to people in various forms.
- Matt gave the Googly principle of simplicity. Strive to make regulations as simple as possible.
- We had many comments from the roundtable audience before and during the session. Carl Malamud gave another principle in a comment: all regulations need to be public, in plain English and make them as accessible and understandable as possible.
Much of the roundtable session focused on what’s currently missing, and what needs to be done next to take these new notions of regulation to the next level. Some of this was big picture, and some more granular.
- Tim thought more work needs to be done on where to draw the new lines between what needs to be regulated and what no longer does because the private sector can better do it, as in the Airbnb example above. Where do we still really need a third-party, neutral government cop?
- How can we better understand the outcomes of government regulation on people? If outcomes need to increasingly be the focus of regulation,then we need to deeply understand them and be able to measure them as much as possible. Tim called for more rigorous academic studies as a practical next step in getting greater buy-in to these new concepts.
- Matt thought we need more “surgical” experiments to show the power of these new approaches. He doubted whether more comprehensive change could come at this point. Better to zero in on some important areas where changes could make a big difference.
- Sophie talked about needing to spread these experiments to more areas of the federal government. Her Treasury Department has been quite innovative in the wake of the Wall Street crash of 2008, but new techniques need to spread to other agencies. There was much talk about the value of experiments at the local city level too.
In the end, the roundtable members had general confidence that these new approaches were getting traction, and there was much potential for innovators inside and outside of government to work together to accelerate changes in the years ahead.
Those in the roundtable demonstrated that potential in real-time. When Sophie pointed out a new Project Catalyst program in new Consumer Financial Protection Bureau that calls for experiments with outside collaborators, Matt responded that he might be able to get engineers at Google to do just that using their famed “20 percent time.” A concrete outcome for this remarkable roundtable.
Peter Leyden Full Post →