(This is a complementary question to: Has the DeWitt Clause, which prevents publication of named DBMS benchmarks, ever been successfully defended in court?)

Several commercial database vendors include an end-user license agreement provision, known as the DeWitt Clause, that prohibits researchers and scientists from explicitly using the names of their systems in academic papers.

There are several concievable approaches to circumvent this prohibition, including:

  1. In the US: Ignoring the prohibition and claiming "fair use" which supersedes the Copyright, or general un-enforcability like Spitzer did (see the related question) - risking legal action.
  2. Outside the US: Some states in the world allow download and certain kinds of non-commercial use of copyrighted works without permission from the author / rights holder, in which case the researcher never accepts the restriction in the first place.
  3. Outside the US: Some states have laws regarding "standard contracts", as opposed to contracts specifically worded by two parties negotiating - e.g. the conditions for opening a bank account, your Internet Access Provider's terms of service etc. Such laws declare certain kinds of terms and conditions in such contracts as null and void if they are injurious to the weaker party - the many individuals who must accept the contract. One could publish a benchmark result, claiming s/he wasn't bound by the DeWitt clause.
  4. Loopholes in the condition itself. For example, Microsoft's SQL server license says:

    BENCHMARK TESTING. You must obtain Microsoft’s prior written approval to disclose to a third party the results of any benchmark test of the software.

    Well, you must obtain Microsoft's prior written approval to disclose to a third party the results of any benchmark. But you're not required to go any significant effort to prevent third parties from learning about these results. So, for example, Joe Smith (not you) runs some benchmark and writes down the results in a file. It's not in a public folder, but - you or a friend of yours have access to that folder. You're not bound by the conditions of the license, and can supposedly publish the benchmark result. Joe Smith should also not be liable, since he did not actually disclose the results to anybody, nor did he solicit others to check out his results.

I'm not saying I know for a fact these avenues of circumvention are bullet-proof. And I'm not asking whether they are. I just want to know whether there are known cases of people using one of them, or any other course of action, to openly and notoriously circumvent DeWitt clauses (in Software or w.r.t. access to materials of academic interest).

  • 1
    I find this very interesting and think it’s mostly on-topic here, but you may get some interesting takes on this subject by asking the same question (or a related question, like what are some hypothetical arguments that may be used to invalidate the Dewitt clause) on law.se.
    – Dan Romik
    Jul 15, 2019 at 17:25
  • A third workaround is to avoid describing the experiments as benchmarks. For instance, "On this website we run horse races, and we found that Maria terminated the race in 17.2 seconds, while Progresso took 23.1 seconds". People have done this to work around restrictions on publishing electoral polls, in my country. Jul 15, 2019 at 17:41
  • 1
    @DanRomik: I might, but note I'm not asking about whether the legal argument is sound, but whether anyone has tried something like that.
    – einpoklum
    Jul 15, 2019 at 18:59
  • @FedericoPoloni: Could be a bit difficult when you want to compare the benchmark results of existing systems and alternative ones...
    – einpoklum
    Jul 15, 2019 at 18:59
  • The legal argument may be sound or not. You can try, of course, to break "contracts". But, I would guess that if you go up against Microsoft or Oracle or similar in court that you'd better have very deep pockets. Very. Deep. Pockets. And I suspect that they utilized their own Very Deep Pockets in the phrasing of such clauses.
    – Buffy
    Jul 15, 2019 at 19:12


You must log in to answer this question.

Browse other questions tagged .