What do valves,
Emergency Department waiting times, MRSA rates, quantum physics and blood
pressure have in common? Slicer tackles the urban myth that measurement is all.
From “his”
earliest recorded times on earth, man has measured things: length, size, number,
duration, seasons, apparent movement of the sun – and it has served him well.
How is it that it has taken him so long to realize the limitations and proper
use of this, one of his primeval tools, however well-polished and refined it
has become?
The Slicer loves valves – not the mechanical kind that
control movement of liquid or gases (useful as those are, they’re a bit dull).
No, Slicer loves the electronic kind, the kind that amplify music, that can be
used to produce the fantastic distortion of a Marshall guitar amp, or a
gorgeous mellifluous Fender sound; and, when properly implemented in hi-fi
amplifiers, can reproduce incredibly natural vocals and acoustic ambience of a
well recorded piece of music, and deliver amazing dynamics. Valve amp design
for these purposes is not an exercise in retro-chic, it’s about conveying the
emotion of the music. The above uses are all far from nerdy. Yet, in order to
get the best out of valves (or tubes as our American cousins refer to them),
the effect of different configurations has to be taken into account.
Unfortunately the different implementations go by terminology that’s
inscrutable to the vast majority of folk who want to listen to music:
single-ended versus push-pull configuration; triode, tetrode or pentode; with
or without transformer output coupling to speakers – and then there is the
multitude of valve types, unfortunately all described by a combination of
letters and numbers.
Slicer’s favourite
valve for hi-fi amps is the 845 – that’s why it’s this website’s pic – 3 of
them, in Stained Glass Window formation. Approaching 8 inches in height, and
originally designed for transmitters, one 845 per channel produces punchy
dynamics and a tighter bass than most other valves used single-ended.
There are solid
state devotees who dogmatically assert that transistors are king, and valves
merely add a euphonic distortion which produces a superficially pleasant warm
colouration at the expense of accuracy and “truth.” Finally there is the
current generation of agnostics or “couldn’t care less” listeners who have long
ago ditched such debates, and traded high quality sound for the convenience and
portability of an iPod.* Those who haven’t opted out of the solid state versus
valve debate often produce “evidence” to support their position. All sorts of
figures get traded: Total Harmonic Distortion, Root Mean Square (RMS) Power,
linearity of frequency response being just 3 examples. Valve amplifiers often
measure poorly with regard to these parameters compared to transistor-powered
devices. Valve advocates have long argued that these measurements are not
indicators of virtue and challenged the measurement fanatics to LISTEN rather
than put their faith in such numbers. In more recent times the limitations of
these traditional measurements have been objectively identified, one example
being that all distortion is not of equal consequence (our hearing mechanism
seems to find even-order harmonic distortion less significant than odd-order
distortion); and solid state devices also produce unintended distortion/interference
in the music signal, but of a kind which did not lend itself to easy
measurement. These findings have provided some vindication for those who
continue to prefer well-implemented valve designs.
In Brownsville Girl Dylan asserted, “People don’t do what they believe in, they
do what’s most convenient…” Scientists are people – they are people in
search of truth, and they/we aim to use objective scientific method to
determine the truth. However, whilst pursuing the scientific holy grail of “the
whole truth and nothing but the truth,” the reality is scientists are sometimes
forced to measure what’s most convenient (what’s most accessible to measure),
rather than what is most pertinent. Slicer has published, elsewhere, on the
difficulties measuring the adequacy of remote, but important, circulations in
the body. In Medicine, even the largest and most obvious circulation – that
through the biggest blood vessels (referred to in an earlier blog as blood
going round and round) - has proven difficult to measure. Doctors for
generations have measured Blood Pressure
at the elbow because it’s relatively easy to measure compared to another other
important descriptor of the circulation, flow. It should be pretty obvious that just because blood pressure is not low, this does not guarantee an
adequate forward flow of oxygen-carrying blood to the tissues of the body. In fact the opposite may be the case –
just try standing firmly on your garden hose as you aim to water some plants –
big pressure build up, but much less water. The analogous situation in the body
is heart failure. Yet, because measurement of flow (output from the heart) has
been difficult, it is relatively rarely performed. It’s not that Blood Pressure
isn’t important, but it’s not the full picture and, when used alone, can be
misleading; if not interpreted correctly, it could prompt incorrect treatment
or inappropriate lack of treatment.
Hospitals are often judged by measuring how long patients have
to wait on Emergency department trolleys before they get admitted to ward beds;
or by the amount of MRSA
(Methicillin-Resistant Staphylococcus Aureus) found in their patients. There is
no doubt that these can be important statistics, but focus on these statistics
at the expense of others which aren’t measured (especially ones which are
important but more difficult to measure) can create a false impression of
reality. For example, there’s not much point getting moved out of an Emergency Department if you’re moved to
a ward which is understaffed or less equipped to deal with your needs than the
Emergency Department; it’s not good to have a target you’re compelled to meet
for trolley waits if the consequence
of meeting it is that you can’t move patients who no longer need to be in ICU
out to wards (because A&E patients get priority of admission to those
wards, driven by a target), and the result is that patients who need emergency
admission to ICU can’t get in because beds are occupied by patients who no
longer need to be there – a result created by excessive emphasis on A&E
trolley waits. Similarly, measurement of MRSA infection rates (on the
questionable and media-driven assumption – because it makes a better story -
that MRSA is a more virulent bug) is of pretty limited value if lots of
patients are dying of MSSA (Methicillin-Sensitive Staphylococcus Aureus), or a
host of other infections, whilst those directed to measure the MRSA rates are
looking the other way. Allowing journalists to set healthcare priorities is a
recipe for disaster. Journalists are often interested not-so-much in the truth
as in what makes a better story.
In quantum mechanics, Heisenberg’s Uncertainty Principle describes how we can’t know all
the characteristics of subatomic particles simultaneously. For example, if we
choose to measure the velocity of an electron very precisely, we can’t have
certainty about its position at that instant. Many view this as a limitation of
our measurement capability eg that in the act of measuring one characteristic,
we interfere with our measurement of another characteristic. The implication of
this would be that if we improved our measurement techniques, we should be able
to measure both precisely. I am informed that this is incorrect – that in fact there is fuzziness about the position of
the electron when we know its velocity accurately – fuzziness that is there in
reality, rather than a consequence of our limitations in measurement. We can know its position accurately, but in
doing so it’s momentum is then fuzzy – not in terms of limitations of
measurement but because, in the subatomic quantum world, a state of definite
momentum and simultaneous definite position doesn’t exist.
Perhaps we are
beginning to understand that the more we know, the less confident we can be
about what can be known, at least by scientific measurement.
“Not everything that counts
can be counted, and not everything that can be counted counts.”
Einstein (attributed).
In the meantime,
let’s get on and enjoy some music…
*Slicer thinks iPods/iPhones are fantastic, but let’s
not pretend they usually carry high resolution music, much less are able to
reproduce it through modest headphones or cheap powered speakers. They are
fantastic because of the lack of moving parts, storage capacity, robustness,
portability and cool factor arising from the artistry of the design – valves by
contrast are HOT, not particularly robust, non-portable and tend to be used in
circuits with dangerously high voltages. Hey, sure danger is what makes loads
of things fun!
OK, Slicer, there is a lot I can agree with there :-) I think your taste in music sounds just fine to me, so no arguments. I have been meaning to get myself a good Marshall and a Telecaster for a while now, ever since I sold my old strat copy and 300W Peavey (which weighed a ton) many years ago following the break-up of "Freudian Slip". At one point we even had a fan, and I am led to believe that Horsey Morgan were singing my song "Then She Smiles" for a number of years after FS folded. In retrospect it remains probably my fave song of that era, but hey.
On to the measurement thingamadoodle - again, I don't have too much difficulty there, but a measurement is pretty pointless without a model to integrate it into, and I think the problems you highlight are primarily *model* problems rather than measurement ones per se. So if I stick a thermometer under the kiddie's armpit and it reads 37.5 that is not telling me that the core temperature is 37.5 - it tells me that thermometer is reading 37.5. I have to integrate that into my model of how kiddies work to gain an *estimate* of what the true temp is likely to be, and with my best guess in place, make a decision.
But measurement itself remains vitally important - measurement extends our senses. It provides us with information that helps us revise and hone our models. If I sequence a gene and get a certain result, I need to combine that with what I know about this gene, genes in general, humans in general and my patient in particular to figure out what it's likely to mean. And I'm a sceptical git - I generally need belts and braces - I don't like results that jar with my overall model - if the picture doesn't make overall sense, then something is wrong - either the measurement or the model itself. I can tolerate some slack in that - there are always unknown quantities - but I think it's important to head for some sort of coherence.
Anyway, we should get together and jam. I'm on a Highway to Hell :-)
Posted by: Shane | 02/22/2011 at 10:39 PM
Oh, I remember hearing about Freudian Slip - sadly never managed to get to hear you guys play. I find it hard to differ with anything you've said here - except perhaps your closing comment (the God you don't believe in hasn't given you up yet). As far as integration into the model goes - absolutely. The problem comes when the relationship between the variable you measure and the one your really want to know is poorer than the example you've given of body temp. And (pertinent to the point we've discussed at length elsewhere) - that thermometer's not much good for measuring blood pressure...
Yeah - we'll have to hook up at some point... we manage to have good chats online - masks off outside cyberland's bound to be good.
Take care
Posted by: Slicer | 02/22/2011 at 11:24 PM