14 December 2009

If you buy a hard disk, and look carefully at the small print on the casing, you will find a number estimating the “Mean Time To Failure” (MTTF) or “Mean Time Between Failures” (MTBF) – to all practical intents and purposes, for the buyer, they are the same thing.

When I first bought a hard disk (to go in an Amstrad PC1512, since you ask; 40 Megabytes, of which the operating system could make use of only 32), MTBF figures were quite small on the scale of things. The mean time before I could expect to hear the heads crashing noisily into the platters and sending all my hard work to the great erasure in the sky was in the credibly near future. Worse still, it was only a mean figure, with no accompanying measure of dispersion; if that conveys nothing to you, don't worry about it ... in essence, the mean figure was useless, since actual failure could happen at any unquantifiable time either side of it1.

Now, as it happens, I have only once had a hard disk die on me before I replaced it on other grounds – usually because I need a bigger one. But the thought of everything vanishing at an arbitrary moment gave me a paranoid devotion to backup régimes, so it seems that MTTF has its uses.

Nowadays, MTTF figures are up in the million hours range ... which means, of course, since a million hours is a little over 114 years, that the disk expects to last longer than I do. It also means that somebody has extrapolated some sophisticated guesswork and we don't really know. And there is still no measure of dispersion, so it's not much use anyway. So, I fall back on empirical experience: I have had hard disks between 100 and 320 Gigabytes in use for some years, and they have never let me down even though some of them have been dropped onto hard surfaces while running. I still run those paranoid backup routines, though: I'm not taking any chances.

This has all come back to mind in a different context.

When Pentax brought out their first digital SLR, the IstD, I shifted my majority photographic practice to it from the LX and K2 film bodies which I had used until then. I have set of *istD bodies, and have stuck to them. New models came and went, ignored.

I would be quite happy to stick with the IstD's forever. Newer bodies have useful new features, and because of what I do I don't have to actually pay for them. But consistency, knowing that every body I pick up is exactly the same, that my hands can operate them unconsciously without distracted my eyes or brain from higher tasks, is more important to me than a new trick or a better bell and whistle.2

Alas, in a world which moves on, it's not always possible to freeze one's own part of it.

Two of my IstD's having bitten the dust3, I automatically looked to the second hand market for replacements. Because most buyers of digital SLRs take a different approach from me, upgrading to new models as they become available, a second hand IstD body now only costs about \100. But the number of available bodies is reducing, and will probably dry up in time. And then again, there is the new spectre of “expected shutter life”.

Expected shutter life is quoted in terms of a number of “actuations”: the number of times the button is pressed. As a crude approximation, this is the number of photographs which the use can expect to take before the shutter assembly fails. So far as I know, this was not given for the IstD; but for its newer models Pentax is quoting an expected shutter life of one hundred thousand actuations; other manufacturers are around the same.

This is very similar in many ways to the MTTF of a hard disk. In particular, it comes with no dispersion measure and so is meaningless as a planning assumption. It may be more realistic than MTTF, in so far as there is no practical barrier to Pentax firing off a test batch of production shutters one hundred thousand times over, let's say, a couple of weeks.

These newer models store within their picture files the number of shutter actuations to date. The IstD did not, so I don't have details for each of my existing bodies. On average, however, I can say with certainty that I passed the one hundred thousand mark for each body about three years ago and have more than doubled it by now. On the basis of normal usage patterns, it's a fair guess that each one of them has passed the dreaded 100K barrier and one or more may well have quadrupled it. So, am I on borrowed time?

The first Pentax body I owned (thanks to the generosity of my parents, who could ill afford it) was a Spotmatic bought in 1968. I still have it, and the shutter still works perfectly in all respects except that the top speed has slipped from the marked 1000th second to a 750th. That slip of the top setting started in 1995, so the camera was 27 years old at the time. I have no way to precisely say how many actuations that represents but, from knowledge of throughput and negative records it must have been somewhere approximately in the region of three hundred thousand. Later bodies have seen similar usage volumes and none of them, where they survive, show any shutter deterioration.

What I want to do is replace both of the lost IstD bodies ... but nagging intimations of mortality urge me to go for a complete newer replacement set. What I've finally settled for is one IstD replacement and one new: a K7 which bears the closest physical similarity to the IstD. I suspect that I have caved into insecurity driven by marketing hype.

  1. If you are interested: the measure of dispersion around the mean is “standard deviation” (SD). If, for example, the MTTF of a disk was given as 100000 hours with SD of 500 hours, it would mean that there was a roughly 70% confidence that failure would occur between 99500 hours and 100500 hours, or about 99% confidence that it would be between and 98500 and 101500 hours.
  2. I have only changed camera bodies three times since 1968. The first time was when Pentax switched lens mount from screw thread to bayonet; to get new lenses which I needed, I had to also get new bodies which would accommodate them – but chose the KX which was very similar to the Spotmatic. The second was addition of LXs; the KXs continued in use, as a set, for work where battery dependency was an issue, but the LX set took over many other types. Finally, the digital IstD replaced the LX while the KX continued (and continues) for film based work.
  3. Curious, the number of euphemisms we have for death and (by anthropomorphic extension) destruction or cessation of function. “Bit the dust”. “Kicked the bucket”. “Popped his clogs”. “Gone west”. “Gone for a Burton”. “Gone to the great [insert own jocular noun here – I used erasure above] in the sky”. “Passed on”. “Shuffled off this mortal coil”. Presumably it reflects superstitious fear of ... well, “The last great adventure”.

No comments: