Remember how back after SOPA ended, the MPAA's Chris Dodd kept going on and on about how he was going to take a more conciliatory and partnership-based approach to the tech industry
(which he mistakenly seems to believe is defined by "Google")? Apparently that's out the window. Today both the MPAA and the RIAA have launched a one-two punch on Google, which is clearly designed to do one thing: get Google to start censoring its search results so that it no longer returns what people are looking for, but instead returns what the MPAA and RIAA think should be
the right search results. The fundamental problem, of course, is that the MPAA and RIAA both seem to think that Google is supposed to deliver the answers they
want the public to see, when everyone else recognizes Google's role is to return the results its users
are searching for.
The "one-two punch" consists of the MPAA releasing a study that blames Google for piracy, followed by the RIAA testifying before Congress that everyone else is bending over backwards to help stop piracy, but Google is failing to do enough. Let's cover these in two separate posts. This one will focus on the MPAA's joke of a study
which does everything it can to pin the blame for piracy on Google, but fails pretty spectacularly as you dig into the details. Amusingly, even the MPAA's own "talking points" on the study (which a friendly bird in Congress passed along) more or less admit that the study is incredibly weak.
The key number, which the MPAA does its best to bury is that search engines "influenced" only 20% of the times when consumers accessed infringing content. This is only slightly higher than the results of our own study
, and it shows that the impact of search on infringement is fairly minimal. Of course, rather than focus on that, the MPAA trumpets a different number: claiming that "74% of consumers surveyed cited using a search engine as a discovery or navigation tool in their initial viewing sessions on sites with infringing content."
Note the careful choice of words: "discovery or navigation tool
." This fits with our own findings (linked above), which found that a very large percentage of the traffic to sites commonly associated with infringement are people simply using Google instead of the address bar
. That is, when we looked at the data, here were the top eight search terms that sent traffic from Google to The Pirate Bay:
- pirate bay
- the pirate bay
- pirates bay
Those are unlikely to be people really "searching" for TPB. Rather, it's just the lazy way of using Google as a "navigation" tool. If that failed, it's highly likely that most of those people would simply try again by using the actual URL. In fact, studies have shown
that tons of people use searches as a navigation tool, rather than as a discovery tool. So, it seems unlikely that forcing, say, Google to return something other than TPB when people search on "pirate bay" is going to prevent infringement. It just leads people to go directly to TPB, rather than using the "short cut" of typing something faster.
Also, the 74% number came from a "survey," and we know how reliable those are, or how they can be heavily influenced by the wording of the question. Hard to take that number as anything particularly enlightening even though it's the one the MPAA is so focused on.
In the talking points the MPAA is passing around, it claims "Search engines bear a huge responsibility for introducing people to infringing content." But that's ridiculous. People know there's infringing content out there. They're not magically discovering it because of search engines. And then there's this:
Take Google's algorithm change as an example -- they held this up last summer as a step that would have meaningful impact and unfortunately we see in the data here that it hasn't.
Um, duh? That, of course, cuts against what the MPAA is arguing elsewhere in this very same release: that Google can have a meaningful impact. Except now they're admitting that it hasn't had an impact even when it did exactly what the MPAA asked it to do
. The point should be clear to pretty much everyone: changing the search algo to favor Hollywood's preferred sites doesn't stop infringement, because people who want to infringe will find a way to do so. It's not that the search engine magically turns people into infringers. The MPAA, not having gotten the results it wants, just wants to do more
algo changing, not recognizing that the end result is going to be the same. There's a word
that describes doing the same thing and expecting different results.
: Matt Schruers digs deeper into the report
and finds out that, indeed, 37% of the "searches" are "navigational" meaning it was just someone typing in the name of the site they wanted rather than doing a search. It also points out that the methodology is that if someone accesses infringing content within 20 minutes
of doing a search, that access is blamed on the search. Dodgy methodology at best.
The report, titled “Understanding the Role of Search in Online Piracy”, asserts that 19.2% of visits to infringing content were “influenced” by a search engine (which MPAA rounds up to 20%, not down, because Hollywood accounting).
What does “influenced” mean? It does not mean, apparently, that the user conducted a search and then clicked on a link presented in the search results. According to the study’s methodology, such an approach would be too “narrow,” thus necessitating a “hybrid” approach that includes any instance where the user visits an infringing site within 20 minutes after conducting potentially incriminating searches. Although the report itself makes no specific policy recommendations, MPAA personnel have complained ,  to officials that search providers’ existing voluntary efforts are no replacement for government regulation.
Less prominently stated is that 37% of searches for infringing content involved a “navigational search.” “Navigational searches” is a term of art (or a euphemism) for typing a domain name into the search bar instead of the navigational bar of the browser. (Most Internet users have probably done this at one point or another, either by accident or out of laziness.) This means that the study’s 19.2% figure disguises the fact that more than a third of the searches that precede a user accessing infringing content via (e.g.) MegaUpload were searches where the us