JimBall,
I agree with your sentiment but not your premise.
Just because there is a fascination with celebrity on PPRune doesn't mean that the professionals don't record or analyse incidents and accidents. The requirement to record is enshrined in ICAO and in a State's regulations. The reported data is used to understand the population of accidents and incidents - the recent thread on the CAA paper "2007-03 - Helicopter Flight in Reduced Visibility" is one example.
The ratio of incidents:serious incidents:accidents is taken from Heinrich's Triangle; he postulated that the ratio is: 330 incidents; to 29 serious incidents; to one accident. It is accepted that if trends can be spotted at at the incident stage, interception strategies can be employed to prevent a serious incident or accident.
As Nick has pointed out, this is exactly the strategy that is used in HOMP; however with HOMP, the aim is to work below the 'incident' threshold by mapping the operational envelope into 'normality' and 'abnormality'. This mapping is not easily done and is the real expertise of HOMP and similar systems. At the simplest level, this mapping provides 'alerts' - which are normally investigated. At the most complex level, analysis using neural net algorithms can spot 'clusters' - a good example of a cluster would be a pilot who always makes a fast approach to a rig; if these (abnormal) approach speeds are below the alert threshold they would not trigger an alert even though the trend might be significant. HOMP data is internal and not reported to the Authority (and sometimes not even to management except in a de-identified form); if an alert (or a series of alerts) comprises an incident, it is expected to be reported by the usual method (MORS) by the pilot.
The CAA hold data that has been reported by the MORS system - that includes some incidents that are below the threshold required to be reported. They are also the recipient of questions from the board of CHIRP - the voluntary and anonymous reporting scheme. (Most questions from CHIRP trigger analysis of the data held by the Safety Data Analysis Unit (or whatever it is called today).)
Authorities (and other organisations) do not just report fatal accidents; for example engine reliability (failures per 100,000 hrs) is required to be examined as part of the Approval for Exposure. A number of organisations collect data for their own safety management systems - an example often used in PPRune is that collected by OGP in production of the yearly oil industry reports; part of this data comes from the HSAC who also collect and analyse data in a continuing bid to improve safety.
Jim