Integrity of the historical weather record.

Scientific measurements over long periods of time may not have high reliability indexes. This should be self-evident and assumed, but in today's digital readouts, computerized analysis and modeling, there is insufficient scepticism in the integrity of data. Indeed, it was recognized when the very first digital readout lab devices began to be used in the late '60s and early '70s, that there was a belief that the reported number was inherently more accurate than a calibrated scale. It took some time to educate technicians that instrument calibration was as important to digital readout devices as it was to metered devices.

As the reports of fudged data surface on DEW (Distant Early Warning) line stations across Canada, the very real problems of data collection are highlighted, as is the reliability of models and prognostications that derive from that data.

Error creep in analysis regimes is a problem which standard tests and techniques may not remove. In fact, the introduction of an assumption to account for a possible error circumstance, regardless of how elegant it is technically or scientifically, still introduces a modicum of error all its own. Back in the early days of mainframe computer programming, we used to call this problem an "error cascade" - it was readily evident in linear programming systems when each error input or calculated compounded the importance of the error, to the point where the results were absolutely meaningless. Many times, the error results were spectacular in their presentation; other times very subtle. Many scientists have had to dine on professional crow when someone pointed out a fault in their analysis path. In part, that is the very nature of peer review in science, and is both a blessing and a curse.

It should be obvious that data analysis dependent on historical records or observations, in the absence of clarity in oversight and methodology, is going to be fraught with error cascades. Oral histories, observations decades or centuries old, will commonly be shrouded in the lint of error. As the DEW line example illustrates, the researcher of the record for which methodologies are published, may not even understand that the error exists. If the anomalous data doesn't deviate too far from the "expected", then its error will likely be understated. Worse - if error is suspected, then the widely divergent data is frequently discarded. Sometimes, that divergent data is valid.

There are significant error cascades in the data packs upon which most of the "global warming/climate change" hysteria is based. Many of the conclusions drawn about anthropogenic global warming are simply not supportable. More worrisome, the crafting of "carbon trading" based public fiscal policy, based on the assumptions in AGW hypotheses, is not only foolhardy, its foolish.

Some years ago, a retailer I worked for part-time when I going to school, used to tell me about the annoying mandatory surveys Statistics Canada ("Statscan") used to send them about retail sales business. The surveys were used to "snapshot" the state of the economy. They took not a small amount of the retailer's time to complete and return, and besides, his view was that his business was none of Statscan's. He would take the form and stuff whatever numbers came into his head at the time, and send it in. Time and time again.