I.E. “The only thing I can’t ascertain is how they handle people like me who almost never watch TV.”

I doubt they really know viewing habits until they contact someone. I ended up having to tell them I don’t watch much TV when they called me in the past asking me to become a Nielsen ‘family member’. I watch very few shows (currently 3, only 1 faithfully, however) but read about what is going on in TV land to satisfy curiosity and in search of new things to check out. I live in a rural area that does NOT get over-the-air with any reliability, and satellite and cable are out because of cost (I do have DSL, though, so I am not completely shut out of things). What I think finally did it was when I told them that (at the time) I was only watching 1 show, a BBC import (Doctor Who) and, when season packs came out on Amazon I got them for Christmas

Haven’t heard from them since lol

And while I cannot comment on the technical aspect of the reports, or what actually happened, I completely agree that Nielsen is by their nature very Soviet about their approach to things. Conservative, tight lipped, resistant to change, suspicious of inquiry, etc.

And what kind of “server bug” would suddenly manifest and shut down their processing for two days? They said it was “firmware”? WTF?

Do they not run test servers before applying patches of any kind to production servers, a standard practice in any well-run IT organization?

This is a VERY fishy statement. This is BS covering for some major screwup, quite possibly involving being hacked.

And, yes, backup systems are common everywhere. You do NOT run servers without backups in any competent IT organization.

I really don’t see how Neilsen can have major hardware issues, at least with regard to collecting and analyzing TV data (they do a lot more than that, of course – they collect retail data as well). They’re only sucking TV data from what, 20,000 sources? Even at 24×7, that’s miniscule amounts of data in today’s terms. Having duplicate backup systems would be easy at today’s hardware prices.

Something went badly wrong when they couldn’t recover in less than three days!

Heads should roll.

Problem is, no one has EVER proven statistically how accurate Nielsen’s numbers are, to any degree. If they have, I challenge you, Bad Robot, to find and publish the proof. Advertisers have agreed to feed the Nielsen Monster for far too long. Nielsen is ASSUMED to be accurate, but they don’t have to bother to prove it because advertisers take them at their word. At first, this was a choice. Now, advertisers and broadcast execs really have no choice but to take Nielsen at their word. Slowly but surely, the STB data collection providers are being purchased as a supplement to Nielsen. Nielsen knows that such a “backup” can take the place of “leader” pretty darn quick! So it’s kind of amusing that Nielsen called a big “sharing” event with the STB data field. That was a post a while back on this very site. I can guarantee that Nielsen has made, and will continue to make, buyout bids on at least one, and probably more, of the STB data leaders. They bought Soundscan when it was proven to be more accurate than their Billboard charts. They won’t hesitate to attempt to buy out TiVo, Rentrak, or TNS either. I kind of hope the STB buyouts don’t succeed .

How long has this SECOND “defect” been around? I think it was an interesting choice of words to clearly lable the first data collection problem as a “bug” and the second thing they noticed when fixing this bug as a “defect”. LIke fixing a broken taillight on the car and finding out that the wiring had been done in the first place.

Oh good. Now we all have tons of confidence in Nielsen. I’m glad they’re not a car manufacturer:

“While failure of important safety equipment on our automobiles is not unusual, the scope of this failure …”

The only thing I can’t ascertain is how they handle people like me who almost never watch TV. I want to feel like we’re ignored in favor of more invested TV watchers (so I have an excuse when I grow attached to a show and it fails), but I don’t have any proof.

lol awesome

If the numbers were that wildly inaccurate advertisers wouldnt pay for them.

How often do things like this happen? Not often. Advertisers would refuse to pay for a duplicate system. Every time someone (including Nielsen with Project Apollo) tries a better more expensive system, advertisers always balk at paying for it, so they fail. If there really was a demand for a better system, advertisers would pay for it or create a new system themselves.

Even for the rational, this major SNAFU highlights why a single-source supplier of any product or service is never a good idea. I work in a manufacturing environment, several plants in several states. We identify backup facilities to support our major customers. If one of our facilities sees an unexpected halt in production (tornado, hurricane, major equipment SNAFU, etc…), then a backup facility will resume essential production ASAP. Nielsen really should, at the very least, have duplicate systems running so that when one crashes, they have a backup plan. “But it would cost too much to do that!!!” Not really – split your current sample in half and run each half through a separate server, maybe?

For the conspiracy nuts: Nielsen would never run split systems because the systems’ results would not be within an allowable range of error, even if the sampled populations consist of identical demographics. One system might say Chuck had 11 million viewers, and the other might calculate that Chuck had only 2 million viewers .

Security of the system? Lemme guess. Someone hacked the system and the data says Chuck pulled a 15.2 in the 18-49 demo Monday night. Despite not actually being on.