Google Book Settlement: More Light Relief

[wolfinthewood]

A consortium of publishers from Austria, Germany, Sweden and Switzerland has put in an Objection to the Google Book Settlement, backed by a number of individual affidavits (for which see The Public Index site).

One point they made I had missed completely, being an English speaker. The parties to the settlement promised the court that they would have the 350-page settlement agreement translated into 35 different languages, and the court so ordered. But they haven't bothered to do this. It hasn't been translated into even one other language, apparently.

More than this: they did translate the Notice of the Class Action into a number of languages: but they apparently used a computer program to generate automatic translations! Which of course were the usual sort of gobbledy-gook. And when you consider that this is a legal document – well.

A German lawyer points out, in an affidavit, a number of critical errors found in the 'translation' into German: 'For instance, the term "sub-class" was translated into a German sociological term - Unterklasse - meaning "lower class" or "working class." ... The term "settlement benefits" was translated into - Sachzuwendungen - an unusual German term used in tax law that refers to a particular form of non-monetary benefits ... Additionally, in the Notice, the concept of "owning rights" was translated into the German legal term Eigentum. This terminology is inapplicable to the Settlement Agreement because it is a technical term referring to ownership of a physical object or piece of land, not the intangible copyrights at issue here.' And so on and so forth.

I found all this hilariously funny.

What conclusions does one draw from this (apart from the fact that the class action requirements for notice have manifestly not been observed)?

A childlike trust in computer algorithms is very Google. And this screw-up - which is a pretty big one - points directly to one of the practical problems with the Google Settlement: that these guys rely far too much on the power of computer programs to transform and manipulate data - and those are just not that accurate. And running a massive publishing enterprise/bookseller/rights registry/library would require pretty accurate data.

Take, for example, the issue of whether a book is 'commercially available', which the settlement proposes to treat as a crucial determinant of the default uses that may be made of a given book in Google's database. And it is Google who gets to decide this: not the Book Rights Registry. Somehow, I don't suppose that Google intends to research that data and hand-enter it on a book by book basis. I bet they have an algorithm for it. And I am sure it is not sufficiently robust for the job.

This is another of the points made by the publishers from Sweden and the German-speaking countries: that Google's database is hugely inaccurate on the question of commercial availability. A South African publisher has made the same point.

While on the subject of Google and stupid mistakes, there is a good post here by Geoffrey Nunberg of the School of Information at Berkeley on the innumerable errors in the metadata on Google Books. Again, some of it is pretty funny: 'Various editions of Jane Eyre are classified as "History," "Governesses," "Love Stories," "Architecture," and "Antiques & Collectibles" ("Reader, I marketed him").' But once again it points to a lack of competence on Google's part. And the long self-exculpatory comment by one of Google's employees frankly doesn't do much rescue them: as Nunberg's interleaved remarks on it politely but firmly make clear.

Then there are the famous one million free e-pub books that Google has recently released. It made good headlines. But someone at Computer Shopper has been trying to actually read them: Google turns classic books into free gibberish eBooks. His conclusion: 'Until Google decides it's going to check the text after it's been through the OCR process, Project Gutenberg remains the best destination for free eBooks.'

And so it goes on.

***

My thanks to the settlement administrator at Rust Consulting, who was kind enough to email me to tell me that they had not only received my email, with the attached scans of my opt out letter, but the letter itself had finally arrived in the post.