In article ,
Duane Bozarth wrote:
Robert Bonomi wrote:

...
The downside to the Sawstop is the _cost_ of an activation. measured both
in time and money, it is non-trivial. circa $80, as I recall, for the
'replacement' cartridge, *plus* whatever damage is done to the blade.

Which would undoubtedly be considerably extended in medical costs and
likely in missed work time, irregardless whether the woodworking is
professional or hobby....

That assumes that the triggering _did_ prevent an accident. grin

Yes, in the case of an _actual_ accident prevention, the expense is
"cheap at {bigmultiple} the price".

In the case of a 'false alarm', it is a totally _unnecessary_ expense.

The trick is differentiating the two cases -- maximizing the former,
and minimizing the latter.

The manufacturer concentrates almost exclusively on the first situation,
and (apparently) totally ignores the latter one.

Which leads the issue of 'false positives' -- those cases where the thing
triggers off for some reason _other_ than a human body-part in contact with
the blade. This is an issue that *NOBODY*KNOWS* how likely it is to occur.
It is impossible to predict what might happen in 'uncontrolled environments
that could trigger the safety device.
...

Well, the rate could be pretty well predicted on the basis of extensive
testing which I would presume they would have quite a bit of...I'm
unaware of them having published any data from which to draw any
conclusions on either side. I would suspect they will have a pretty
good idea before they commit to production, however.

Obviously you're not aware that the saw *IS* in production. grin
They've been delivering since last fall.

And that "lack of published data" is _precisely_ the point. Emphasis on
the word "PUBLISHED". If the manufacturer knows, they're *not*talking*.
Which leads one to ask "why _not_?"

I can think of only _two_ possible answers to that --
1) they do *not* have comprehensive false-triggering data.
2) the data shows an 'unacceptably high' rate of false-triggering,
and disclosing it would adversely affect their marketing.

I do *NOT* have any reason to believe that #2 is the case.

I strongly suspect that #1 -is- true. It is *very* difficult to test for
'unexpected' circumstances. It may seem trite, but if you can think of
it happening and test for it, then it is _not_, by definition, an 'unexpected'
situation.

One kind of a "silly" example:

You're making a zero-clearance insert, from some plastic 'scraps' obtained
from a local manufacturer. You trim to size, put it in the table, turn on
the saw, and start to raise the blade.

*BANG*

It turns out that that piece of plastic was sufficiently *conductive* to
trigger the protective mechanism.

_Could_ that happen? *You*betcha*! How likely is it? *GOOD* question! I
don't have the data to begin making an estimate.

Is there any _rational_ way for the manufacturer to _test_ for it?
And, if they do, what does it show?


There is a saying in the Q.A business:
"For every fool-proof system there exists a
*sufficiently*determined* fool capable of breaking it."

*NOTHING* can substitute for a few million hours of actual use by the afore-
mentioned "sufficiently determined" types.

"Discovered bugs, are finite in number. *UNDISCOVERED* bugs, on the other
hand, are, by definition. _infinite_ in number."