In a controversy that has doubtless become the subject of conversation in most academic circles lately, Google has been accused of paying academics for research supporting its policy interests.
Readers should note that while this controversy has been picked up and reproduced on several news reports, the primary allegations were mounted by an organization called the Campaign for Accountability (CFA) which issued the report earlier this month.
The Wall Street Journal (WSJ) then picked up on this and echoed most of the findings.
Campaign for Accountability: Introductions
The ‘Campaign for Accountability’ (CfA) is a “non-profit, watchdog organization” with a research initiative called the (GTP) ‘Google Transparency Project’. According to its website, this initiative aims at bringing transparency into Google’s business and functioning. There are various projects under GTP, which can be found on its website. The initiative at the root of our controversy is one titled, ‘Google Academics Inc.’
The ‘Google Academics Inc’ report (which can be downloaded in its entirety here) alleges, “Google has exercised an increasingly pernicious influence on academic research, paying millions of dollars each year to academics and scholars who produce papers that support its business and policy goals. The academic papers examined encompassed a wide range of policy and legal issues of critical importance to Google’s bottom line, including antitrust, privacy, net neutrality, search neutrality, patents and copyright. The number of Google-funded studies tended to spike during moments when its business model came under threat from regulators—or when the company had opportunities to push for regulations on its competitors.”
The report also names several academics who have allegedly been the drivers of Google’s “corrupting influence on academic research.” The allegations are hinged on an apparent lack of disclosure of Google’s funding by the academics so named.
In its report, CfA includes a list of papers that have been funded by Google both directly and indirectly:
|Indirect Funding||Direct Funding||Total|
|Funding not acknowledged||130||85||215|
It further says, “In more than half of those cases (54%), academics were directly funded by Google. The remainder worked for, or were affiliated with, groups or institutions that were funded by Google. In the majority of cases, readers of the papers would not have been aware of the corporate funding: Academics did not disclose the Google funding in nearly two-thirds of cases (65%). Authors failed to disclose funding even when they were directly funded by Google in more than a quarter (26%) of cases.”
The allegations become increasingly vicious through the course of the full report. They include inter alia non-disclosure of funding, submission of drafts to Google prior to publication, lobbying for google etc.
To be able to clearly understand how CfA came to the results stipulated above (and also in the interest of being true academics ourselves) we must take a few minutes to peruse the methodology adopted by CfA. The methodology page opens with the submission that the project combines “anecdotal inquiries into conference proceedings of policy issues of interest to Google and curricula vitae of authors with known ties to the company” with structured searches of Google Scholar – the search engine for published, academic work.
The structured search methodology can be summarized as –
Structured search number 1 (SS1):-
Step 1: Search Google’s disclosure pages for recipients of direct research support.
Step 2: Search these names on Google Scholar in conjunction with policy keywords such as copyright, patent, net-neutrality, anti-trust, etc.
Step 3: Filter results for papers published after the date of disclosed funding.
Step 4: Manually filter for author identification and Google support acknowledgement, etc.
Structured Search number 2 (SS2):-
Step 1: From SS1, pick out papers that explicitly disclose funding.
Step 2: Search Google Scholar for instances when policy-keywords from SS1 coincides with acknowledgement languages such as “google fellowship”, “thanks google”, etc.
Step 3: Filter for verification of acknowledgement of Google support, etc.
Structured Search number 3 (SS3):-
Step 1: Compile list of authors in anecdotal research and SS2 and search their names + policy keywords on Google Scholar.
Step 2: Filter for papers that acknowledge Google support in one publication but don’t in others – all being of the same topic.
Step 3: Filter to verify if paper was published after date of first known Google support, and support acknowledgement, etc.
My problems with this methodology can be explained through a few illustrations:
Say, Google’s disclosure pages (Not sure what exactly these are and how specific the disclosure is for each case. I accessed a list of public disclosures here, but they are hardly research/academic specific) reveal that it has paid $10 for research by A in Q1 of 2010, $10 to B as part of its fellowship, $10 to a third party organization by which C is employed.
If we follow the methods prescribed by SS1 and SS2, all papers written by A, B, C and D after the years mentioned (and those with subjects coincidental to Google’s policy interests) are under the scanner. Despite the possible lack of uniform disclosure by Google, and also despite the fact that some of the papers could have been written independent of funding.
SS3 supposedly helps in collating the names of authors who have disclosed funding in one paper, but not in another – both being of the same topic. If we go back to step 1 of SS3, it is clear that this search begins with the names of authors derived from SS2 and “anecdotal research”. So, effectively (and if I understand this correctly), while SS2 gives you authors who have explicitly disclosed funding, the “anecdotal evidence” is used to confirm for papers of similar subject matter where authors have not disclosed the same. This seems a little arbitrary because there is no verification of the exact scope of the funding dispensed. Instead, it categorizes all papers of similar subject matter as having received funding and not disclosed the same. This is an indicator of CfA’s tendency to make such a blanket generalization about funding of papers – implying that its conclusions follow the debatable assumption that, once a Google funding recipient, always a funding recipient.
Another instance where this assumption takes effect in CfA’s report, and more worryingly so, is in the case of “indirect funding” of the papers. CfA does not expressly state what this term means, but out-rightly attacks several professors for benefitting from such indirect funding. Apart from “anecdotal evidence”, other evidence used in the searches for “indirect funding” include inter alia email correspondence, attendance in conferences funded by Google, linkages to organizations that have in some way been or are connected with/funded by Google.
Say, emails reveal that author ‘A’ “thanked” Google for “support” and also submitted pre-published drafts of a paper. Author B has been linked with Google in the past, and has authored several papers that are similar to Google’s policy interests. Author C has been to conferences that have been directly or indirectly funded by Google. As per my reading of the report, CfA considers A, B and C as academics under the “indirect funding” category. It does not consider that such a classification might be flawed in that the link between Google’s so called funding and the publication of the paper, might be too remote. Or alternatively put, that B or C might have written papers without actually being funded to do so.
This categorization begs the question: Does any sort of prior linkage with Google taint all subsequent research as obtained through undisclosed, indirect funding? Or perhaps the question should be – does any sort of linkage with Google taint subsequent research as biased from the outset?
That, I believe, takes us beyond a mere analysis of this report into a reflection of the ethical norms surrounding our academic ecosystem. But, let’s hold that thought for a moment. Speaking objectively of the methodology adopted for this report, my point is that there may be some amount of conjecture to all of this.
The CfA report: What do academics have to say?
Similar reservations about the conjecture underlining the CfA report have been voiced by academics as well. An op-ed by Annemarie Bridy and Aaron Perzanowski (both, scholars who have never received funding from Google but have been so named in the report) published in the LA Times debunks the report, primarily attacking its “shoddy methodology“. To quote the scholars’ weighty counter, “Once a scholar was “Google-funded,” all of his or her subsequent academic work, even work undertaken at a new institution, was tainted in the Google Transparency Project’s reckoning…the financial connections to Google for many academics included in the Google Transparency Project’s database are tenuous at best and sometimes verifiably nonexistent. Dozens of scholars already have objected to the inaccurate inclusion of their work.” The op-ed also notes that while CfA points fingers at the lack of transparency and openness in Google’s funding of academics, the CfA itself is terribly opaque and has failed to disclose its source of funding.*
Another coverage by Wired available here, outlines the holes in the CfA report, naming more such academics who were unfairly mentioned.
It is interesting to note that since the publication of the report, the CfA has also published an “update” in response to the many complaints it received regarding its report. Some of the concerns have been addressed adequately, most have, however, been dismissed in the interest of public knowledge of Google’s “corrupting influence”. Readers can be their own judge here.
Industry funding: much ado about..?
Despite all the criticism of the methodology, it would not be wise to stray from the primary point of the CfA report: that we may be up against undisclosed, industry sponsored research. The question is one of balance and there is no clear black v. white box here. As I mentioned above, we may be up for some serious reflection as to how we should deal with industry funding. So what ought the norms to be?
Last year, we had both Prashant and Prof. Basheer give us their opinions on the subject. In his posts, here and here, Prof. Basheer argues that industry funding does raise a prima facie presumption of bias, given that in most cases industry pays for a purpose and to push an agenda. But, merely because something is funded by the industry, does not necessarily mean that it is biased. One has to look at a whole host of other factors such as whether the funding has been disclosed, did the funder influence the outcome of the research, the academics’ past record, etc.
The one key norm that should be promoted by all is “transparency” and academics must always take care to ensure that the source of funding is disclosed. Even if they have to err on the side of caution and disclose in big fonts and letters, and in more places than one!
Suggestions for transparency can also be found in the Open Letter on Ethical Norms in IP Scholarship that aspires to bring high standards of ethics within IP academia. It too, inter alia, admits that industry funding is likely to produce biased results and suggests disclosure, peer-reviews, open inquiry, etc to ensure transparency.
Prashant’s published views on the matter are restricted to industry-sponsored conferences, and throw up a contrast. He questions the assumption that industry sponsored conferences are almost always biased towards certain kinds of outcomes. He also suggests that on some level, such funded conferences might be beneficial for Indian academics who routinely suffer lack of funding or effective networking opportunities.
The LA Times op-ed cited in the previous section harps on a similar perspective, “Corporate funding alone does not mean that research is unreliable. As federal and state dollars available for research continue to dwindle, more research is likely to be funded directly or indirectly by corporations, foundations and other private sources…But ultimately, regardless of whether research is privately funded, its results should be evaluated on the merits.”
The IPKat coverage of the controversy offers another interesting opinion. Regarding sharing of drafts with corporates, the author notes that this is a common practice in order to incorporate practical, non-academic expertise into academic work. She notes, “This kind of peer review, when it does not cross the line into editorial control, leads to better research more grounded in reality.”
Industry funding: an ethical question
It seems to be clear that whatever be your personal views on this issue, industry funding is here to stay. In fact, the Wired report cited earlier notes a significant comment by CfA, “The larger issue is Google’s attempt to influence all these academics.”
The issue is one of ethics. And while we try to figure out how best to work with industry funding, I humbly contribute some questions that came to me while attempting to cover this controversy:
- Should industry funding of research be treated subjectively – depending on whether or not the funding is a long term fellowship or a short term project? If the funding is more long term, should all the academic pieces by funded scholars even in much later years be attributed to the funding and disclosed?
- Should we have the same disclosure and other norms for academics who hold certain views and then are approached by industry and funded? I.e. if they endorse the same views to begin with?
- Since a large chunk of the CfA allegations were based on treatment of “biased research” by regulators – Should there be norms outlining what kind of research govt regulators can or ought to rely on when making policies. How much weight should they give to industry funded reports?
I’d like to conclude by going back to CfA’s (rather noble, if I might add) objective – to bring accountability to Google’s doorstep. Google is a booming corporate which has a pervasive influence on our public and personal lives. Its policies are undeniably important to all of us – be it privacy (especially so for India at the moment), anti-trust, net-neutrality or copyrights. At the very least, CfA prompts an important conversation.
* Information surfaced sometime last year that Google Transparency Project is being partly funded by Oracle, see more here.
Image from here.