The Case for the Digital Platform Act: Market Structure and Regulation of Digital Platforms. By Harold Feld. Roosevelt Institute & Public Knowledge; 216 pages.
Public interest advocate Harold Feld has a new book on tech regulation, which means we all need to stop what we’re doing and take note. In “The Case for the Digital Platform Act,” Feld lays the foundation for a major regulatory intervention in the tech platform space. We’re talking Amazon, Facebook, and Google.
Jointly released by Public Knowledge and the Roosevelt Institute, and with a foreword by former FCC Chair Tom Wheeler, the book arrives at a time when Senator Elizabeth Warren is calling for structural separation—aka, a breakup—of large tech platforms plus a non-discrimination regime to police large and small tech platforms alike.
On Warren’s tech breakup proposal, Feld begrudgingly concurs with her plan, despite “enormous implementation problems,” arguing that “Congress should certainly authorize structural separation where necessary to further competition or the other goals of the statute. But it should carefully define the triggering conditions.” Feld never defines the triggering conditions precisely, perhaps in recognition of the complexity of this undertaking. He prefers product unbundling as a pre-condition to structural separation, pointing to the success of unbundling in the case of Internet browsers.
Among a host of regulatory tools for policing the tech platforms, including data portability and privacy, Feld lists “non-discrimination rules” as a priority for legislators. Because that topic is of acute interest to this economist, I focus on non-discrimination here, leaving the reader to explore Feld’s prescriptions for other tech-related competition problems.
Feld cites the program-carriage rules from section 616 of the Cable Act as one historical precedent for a non-discrimination regime. But he is generally negative on its efficacy, saying that “Section 616 has generally proven ineffective. Independent programming has remained negligible.” He is too harsh on both counts.
Two independent cable networks, the NFL Network and MASN (jointly owned by the Baltimore Orioles and Washington Nationals), have secured relief from discrimination as settlements to section 616 litigation, in the form of broad carriage on Comcast. And two other independent networks, Tennis Channel and GSN, secured findings of discrimination by the FCC’s administrative law judge against Comcast and Cablevision, respectively, only to be overturned by a conservative judiciary (Tennis) or Commission (GSN). [Disclaimer: I was the economic expert for the complaining networks in these matters.] This is hardly a record of futility. In the late 1980s, before the advent of these protections, there were fewer than 40 independent cable networks; today, there are over 300, which is hard to square with Feld’s characterization of a “negligible” contribution from independents. In fairness, it is impossible to know how many independent networks there would have been in the absence of the rules, but I note that independent networks have grown at a faster rate than cable-owned networks since 1992, suggesting that the program-carriage rules had some positive impact on edge innovation.
For the uninitiated reader, Feld helpfully spells out the difference between non-discrimination and common carriage, explaining how the latter is not applicable in the digital arena:
Common carriage requires the service to treat all similarly situated customers the same and prohibits any unjust or unreasonable discrimination among users. For some communications-like services, such as messaging, common carriage may be appropriate. But an important role of many platforms, such as search, is to help users sort information. This makes common carriage largely inapplicable to digital platforms. But other forms of non-discrimination, such as a prohibition on favoring affiliated services, are both feasible and in many cases appropriate.
Feld correctly recognizes that in the absence of any non-discrimination rules, tech platforms will favor their affiliated content (in the case of Google) and merchandise (in the case of Amazon), potentially undermining innovation at the edge of those platforms. Without such protections, the tech platforms could also “capture new related markets, or extort rents [from independents] in exchange for access.”
In November 2017, Senator Al Franken wrote that “As tech giants become a new kind of internet gatekeeper, I believe the same basic principles of net neutrality should apply here.” Similarly invoking the spirit of net neutrality, Feld explains “Additionally, as in telecommunication, nondiscrimination and neutrality are important tools to ensure competition on the provider side of the two-sided market—even in the presence of competition between platforms.”
[Another disclaimer: Feld and I sparred for the first time on net neutrality at George Washington University in 2007, three years before the Federal Communications Commission first implemented the Open Internet Order in 2010. Although Feld and I are converging on the major policy prescriptions—bright-line prohibitions on blocking/throttling, and a standard to police mild forms of discrimination such as zero-rating—Congress doesn’t appear any closer to resolving the issue of legal authority.]
In addition to economic concerns, Feld notes that another “reason to mandate some form of nondiscrimination [for tech platforms] may be to prevent manipulation of news or suppression of speech.” Despite the need to police platform discrimination, he recognizes that platforms such as Google “must inherently ‘discriminate’ to perform their functions.” Thus, a rule that barred discrimination narrowly on the basis of affiliation would permit a platform such as Google to discriminate in search results based on (say) the match between a user’s request and the best information available on the web. Even Google's precious OneBox would be spared!
In terms of a crisp policy prescriptions, Feld proposes a rule that would “[l]imit nondiscrimination to harmful economic discrimination.” This makes sense, and it is consistent with the evidentiary requirement imposed on complainants in program-carriage cases to show that the discrimination “unreasonably restrained” their ability to compete effectively. Regarding other evidentiary burdens, Feld notes correctly that “a harmed party does not need to prove discriminatory intent or other improper motive;” evidence of competitive effects should suffice.
A common critique of any behavioral remedy is litigation expense: Can the independent afford to duke it out with the platform for several years before relief is granted? Feld notes that “arbitrating complaints by rivals that the dominant firm is favoring its affiliate can be lengthy and time consuming.” To address this problem, he calls for “mandatory timelines to resolve complaints coupled with clear appeal rights.” In particular, Feld says the relevant agency should be given 60 days to dismiss the complaint or refer the case for further action. He also calls for a private right of action, to ensure that enforcement levels do not ebb due to a “lack of political will.”
To ease litigation expenses, Feld cleverly suggests that complaints could ask enforcement agencies to test algorithms for prohibited bias “through a ‘black box’ process that shields the code from repeated testing designed to reverse-engineer the algorithm.” Because the tech platforms will often possess the data needed to prove discrimination, Feld calls for a burden-shifting regime, in which a complainant need only make a prima facie case, at which point the burden would shift back to the platform. It bears noting that this is effectively the technique used to adjudicate program-carriage disputes.
The Case for the Digital Platform Act notably excludes Internet service providers (ISPs), as well as operating systems and other platforms, from its definition of “digital platforms,” leaving open the possibility that ISPs would be regulated pursuant to a different non-discrimination standard. Could Feld live with a single, symmetric non-discrimination standard that would police tech platforms and ISPs alike? Clearly, he and Public Knowledge still want the 2015 net neutrality rules for ISPs, and his book notes that common carriage (in contrast to non-discrimination) would not work for tech platforms.
But what if a single non-discrimination standard could achieve more political support, obviate the need for net neutrality legislation, and possibly diffuse ISP opposition? I put this question to Feld, but he wouldn’t bite: “I always start with the ideal, rather than start with the ‘politically feasible.’ The risk of pricing yourself out of the market is much less than the risk of shortchanging yourself.”
You’d be shortchanging yourself by not reading the book of such a principled advocate.
Twitter: @halsinger
