RGS65 wrote:
What Kirk said is of course correct (and he would know!) Of course Canon has this data, and more likely they have tested it themselves, just like every other product manufacturer. Not only that, but I'm sure the engineering isn't all that different shutter to shutter across the lines so the experience from other cameras is also relevant. The warranty isn't based on a guess - its analyzed and part of the cost estimates the company has figured in to warranty repairs going forward.
Right, of course they tested it themselves. I work in software QA and we run hundreds of thousands of iterations of end-user scenarios on the products before they go into production. Even doing so, there's no way in hell we're going to assume we've caught all the issues that will be found in production. You take those hundreds of thousands of iterations, then multiply that by hundreds of thousands or millions when the product goes gold, and every little failure gets ripped out of their little nooks.
Large-scale software products are a bit more complex than cameras in that there are generally more ways they can "break," but I'm trying to assert that Canon does not have reliable data on their cameras prior to releasing their product to production. They cannot test the products nearly enough, or in nearly as many ways as the consumers will. Even long after the product has come out, with an item like a camera there's a ton of variables involved in gathering MTTF data. Statistics based on small numbers can only go so far. It's not as difficult in software, particularly in software that's coupled with the internet where a huge amount of data can be mined.
Large-scale software products are a bit more complex than cameras in that there are generally more ways they can "break," but I'm trying to assert that Canon does not have reliable data on their cameras prior to releasing their product to production.
Of course, there will be only a limited number of test lab conditions and a limited number of test cases a company can run. A few hundred thousand users may represent at least a thousand different conditions and a million test cases.
So I would agree that it's certainly possible Canon is increasing its shutter life estimates based purely on "real world" data that the shutter design is doing better than they expected.
While I agree that some shutters fail prematurely, and that a shutter + sensor replacement is expensive, unless one hits the said rating of a camera, replacing the shutter as PM is still not supported statistically. The failures are just too far in between stable cameras.
Jeremy, my friend, I can't say much about your luck
