ComScore has been rolling out the new system for months and says it can now use it to report on 25 percent of the 50 biggest sites on the Web. Another 50 percent of the top sites have agreed to work with the system, Abraham says.

ComScore lets publishers who are already clients participate in the program for free. But it will charge everyone else $10,000 a year, which the company says helps cover the cost of new servers and other equipment it needs to process the new deluge of data.

Gawker/Valleywag says it more bluntly “Comscore Blackmail: Pay Us $10,000 or We’ll Keep Under-Reporting Your Traffic”. Because the figures are consistently biased downwards in the legacy reporting, it creates a distinction in traffic reporting between those who place the tracking codes on their site and those who don’t, that is very correlated with higher ad revenues today (hence the idea of “blackmail”). Most sites won’t ever pay $10,000 so this means only the largest sites will even consider doing it, lessening the burden for comScore’s (probably not very efficient) infrastructure for doing so. This distinction in the costs to participating vs. not exists in a very different way with Quantcast, although the compensation and benefits a publisher gives up are a bit more subtle in their case since there is no cost for participating with them – namely that Quantcast will use their aggregate data to help marketers find similar audiences. By no means a small or easy task and kudos to Quantcast for taking that on. Reporting is one thing – but actionable data based on reporting is quite another.

Three points worth making, though this is interesting and there are many more to think about:

1) the “size = more ad revenue” should (but will take some time still to) become less meaningful, as more centralized inventory access and price discovery is possible. A good reason to work with three big sites versus thirty small- or medium-sized sites is that trafficking and setting up ad buys with people takes time and is grossly inefficient. And we’re all nothing if not lazy in the media business.

2) there is no doubt that the panel approach undercounts data, especially for the at-work and university-type audiences amongst others. I worked at Nielsen NetRatings and the most common argument we used if we ever gave any credence to these arguments was that most sites were consistently underreported. Hmmm yes. Of course. So it begs the question – if you are being trusted as someone to measure, report on and (often) opine on the health, status and metrics of an industry, don’t you want to be as good as possible which in turn leads to more trust, more business and more money? It is unclear in this case.

3) The capacity and desire for sites to implement a lot of different companies’ javascript code on their site is limited. Sites don’t want to break if any of these third-party pieces is messed up. Use firebug/firecookie or Charles and see what all is loading on the typical large site. Risks increase, and the infrastructure capacity of many of these vendors is far less tested than you might think, especially if the publisher is now being asked to pay for the infrastructure (another typical argument used by NetRatings or comScore or others “the market needs to pay high prices to help fund good quality measurement” which is also total BS).