Saturday, February 02, 2013

Google funds French publishers

Google signs a deal with the French government to provide 60 million Euros to establish a fund to finance projects by French publishers to improve their Internet presence.

It sounds a little bit crazy that the European publishers have to beg a competitor like Google to fund the innovations they desperately need to survive.
Google will have total visibility of the project developments, and if any relevant and innovative result would come out of the projects they will be in a position to exploit it, possibly even more quickly than the publishers, since Google has a much bigger and better infrastructure and a larger and well trained technical staff.

History teaches that the typical effect of subsidies like these is that companies will just relax on the little oxigen for their finances, but when the money will expire, they will be in even worse position than before in the marketplace.

The second part of Google proposal is to increase the use by publishers of Google advertising platform which guarantees to Google a 62% share of the revenues.
All things considered, accepting Google proposal would be a suicide for publishers, as argued by the director of Fanpage.

Even more striking is that Europe is unable to fund its own research: it is a terrible sign if private US companies start replacing the ineffective and bureocratic process of funding research established at the European Commission.
When I was at MIT, Marvin Minsky impressed me by saying that it is much better to fund research through mecenatism than by committees. Even random choice is better that committees since there is a chance that an outstanding project will be chosen, which instead would have zero chances in an evaluation by a committee.

After having been one of the pioneers of European projects back in 1980's, when one of my project was even selected as a ESPRIT flagship project, in the 1990's I started experiencing the miopy of the European research evaluation process.

Specifically on the topic of technologies for newspaper publishing, in 2011 I gathered a team of the best world-reknown experts in Europe on syntactic analysis of natural languages and submitted to a call of FP7 a proposal for a project titled
NewsFlow: An Integrated Language Engine and Personalised Newspaper Delivery
The project aimed on one side at advancing language analytics technology, and on the other to apply these technologies in the production of personalized newspaper for online fruition.
We envisaged building an innovative service for online journalism, capable of analyzing and aggregating news-items on a large scale, and allowing publishing personalized newspaper editions and delivering them through different digital media.
The project involved the largest newspaper publishing group in Italy and had preliminary contacts with other European groups.

The reviewers of project NewsFlow (proposal n. 287731, call FP7-ICT-2011-7) seemed quite incoherent to me as they wrote that the project had high potential impact:
"The objectives are clear and valid. The goal of building a single all-encompassing model is innovative and ambitious but risky. The overall concept however is sound and is in line with the work programme."
"There is high potential impact, but this is jeopardised by the risks embodied in the technical approach."
but then suggested not to fund it.
The project would be well under way by now, if it were funded by the EU and some of those technologies that publishers seem to require now would be already available to them.

The fact that the project was ambitious but risky to me was a very positive assessment.
If one does not undertake risky projects, what is the purpose of research? Do just what you already know how to do?
Isn't the purpose of public funding to be used for ambitious and risky ideas?
If it is not risky, one would ask a bank for funding.
Should the EU Commission reason and behave like a bank?
Whose risk it was? Did the reviewers worry about their own money?
Did they ever worry about the many funded projects that have not delivered anything, even if the project was not risky?
What is more risky for Europe: not to try innovative ideas or to wait for some US company to take their market by storm?

An internal review document by the Commission itself acknowledges the issue but does not propose a solution: "Challenges: The Need to Reach an Agreement on the Balance between Trust and Risk in Research Funding".

I am pursuing nevertheless the idea in which all the NewsFlow project proposers believed and we are working on the risky "single all-encompassing model" with various funding, including, ironically, a Google award.

This is the second time that the Language Technology division of the European Commission rejects a proposal of mine.
The first one was in 1996, when I submitted project Naxos (ESPRIT Proposal No. 24872), for developing search technologies based on link analysis of Web (before PageRank was made popular by Google).
In that case the project was rejected with the motivation it "lacks a European dimension".
The European project officer at the time, Nino Varile, was unable to explain to me what that meant.
The project fullfiled the requirements of participation of teams from several European countries.
We then built those technologies ourselves with our own startup, Ideare, who went on to built Web search engines throughout Europe (lack of European dimension!).

No comments: