Readers of Media Week saw this blind side in the mid-80s when I wrote a rather thuggish piece criticising one of the JICs for the volatility of its data.
Simple soul, it seemed to me that research should be a photograph of what's going on out there, so changing the methodology should be like just using a different lens; different texture perhaps, but still identifiably the same picture. Apparently not.
Researchers confidently expect any change of tools to produce unrecognisable numbers and a revalued currency.
But there's a problem here. As people's media usage splits into new directions, old research techniques and standard sample sizes struggle to keep pace; so the price of research reliability for each media sector rises, just as its revenues are fragmenting.
It's in this context that Rajar has just announced an increase of a third in its research outlay to £6m in 2007. Its statement last week promised "stable, robust and recent" data, meaning it has decided not to trade in the trusty old diary for the Joe 90-type wrist-watches that the amusing Kelvin MacKenzie tried to bludgeon the industry into adopting during his brief but noisy radio life.
The new investment buys an upgrade of the diary to encompass new platforms alongside a joint test with Barb of people meters - devices that hang round the necks of volunteers, listening in to their radio and TV usage.
But, mercifully, Rajar has not been seduced by the obvious grooviness of these must-have fashion accessories into abandoning the diary, so we will be spared the unrecognisable conclusions and fuzzy researcher-type excuses. Just to add gold taps to the whole thing, it has hired another company to improve sampling, weighting and other recherche stuff.
It seems a good - if expensive - balance; one that other media sectors will have to find.
New technology has a part to play in tracking new behaviours, but no economy can function right if its currency bounces around like Uganda's.
- Richard Eyre is chairman of Atlantic FM, a new radio station for Cornwall, email@example.com.