EVNow
Well-Known Member
OT :
There seems to be a couple of ways this can be measured. First we talk about error rates - and reliability is the reverse of that. So, if we have an error rate of 0.1%, then reliability becomes 99.9%. The error rate can be expressed in terms of
- length (as in miles, kms, millions of miles etc) : common for vehicles
- time (hours, thousands of hours, years etc) : common for electronic products
Then, there is the nebulous aspect of % of scenarios covered. This makes little sense, since the # of scenarios is not finite.
So, it looks like reliability will be a confusing metric, just like EV Range.
ps : When you see error rates of car driving, even within per million miles usage you will see different metrics as well. Most often quoted is fatality - and as EM has talked about, this is (fortunately) too low for establishing FSD being better. Then there is the # of injuries per MM. But the one I'd use is # of crashes per MM, as that is what FSD needs to avoid.
Yes, unfortunately there doesn't seem to be any standard way to talk about this. When people claim 99.99% (reliability) they don't say how that is being measured.Let's take the data above. In the prior post we learned that the Google study estimated 4.2 crashes per (1E6 mi * 5280 feet/mi * 12"/ft) = 7 billion inches! 99.999997%!!!! Level 5 here we come!!!
There seems to be a couple of ways this can be measured. First we talk about error rates - and reliability is the reverse of that. So, if we have an error rate of 0.1%, then reliability becomes 99.9%. The error rate can be expressed in terms of
- length (as in miles, kms, millions of miles etc) : common for vehicles
- time (hours, thousands of hours, years etc) : common for electronic products
Then, there is the nebulous aspect of % of scenarios covered. This makes little sense, since the # of scenarios is not finite.
So, it looks like reliability will be a confusing metric, just like EV Range.
ps : When you see error rates of car driving, even within per million miles usage you will see different metrics as well. Most often quoted is fatality - and as EM has talked about, this is (fortunately) too low for establishing FSD being better. Then there is the # of injuries per MM. But the one I'd use is # of crashes per MM, as that is what FSD needs to avoid.
Last edited: