Psybertron Asks

Scientism is a problem, but this is scary – scientists being encouraged to date other scientists.

I read this Ben Goldacre piece a couple of weeks ago. The problem one always has to ask is … is this kind of bad science accidental or in some sense deliberate – a skilled incompetence either by the practitioners or their managers / editors / reviewers, or both in a kind of tacit collusion. In situations of complex human endeavours some hypocrisy is inevitable, to balance motives and goods across multiple levels, and a degree of trust is therefore also inescapable. Science is no different, taken as a whole “business”.

Personally, I’m more against bad scientism, using science badly in situations that are far from scientific – rather than good or bad science per se. With infinite time and resources you could argue all situations can be reduced to science, but the reduction can discard the real world value. Statistics is of course one of those techniques used to bring the vagaries of human behaviour into the scientific space in quantifiable chunks. This adds another level of complexity to the whole exercise leading to more possibilities of evaluating the wrong things, and/or evaluating them wrongly.

Ben’s story above is about the statistical methods, this story today in The Scholarly Kitchen (via David Gurteen and Stephen Downes) is about choosing the wrong inputs for the wrong motives – citations, again. Proves the point that science is a messy business, parts of which are far from scientific.

And of course, the “Measuring the Wrong Things” headline is one a long line including Einstein’s “Not everything that counts can be counted.”

George, can you provide example(s) of where “eulogies” for Chris Hitchens also attacked you ?

Interesting after the piece earlier via David Gurteen on subjectively assessed management metrics to hear this W Edwards Deming quote again (also via David)

Eliminate numerical quotas,
including Management by Objectives.

I blogged this link the day the story came out, to Facebook and/or Linkedin, but of course that doesn’t preserve it in my database, so I’m repeating it here. What is really scary is not the persistent pilot error: The inexperienced co-pilot may well have been disorientated or even in some kind of “personal mental autopilot” denial as to the true state of the aircraft despite clear and specific audible and verbal warnings. Schoolboy error to pull the stick back under those conditions, let alone a qualified pilot. I would say it must count as a design fault in the A330 (and presumably all the current generation Airbuses) that the crew do not get any direct feel or instrumented feedback of the control surfaces. The experience in the cockpit then counts (counted) for nothing. How is that “averaging” stick behaviour design rationalised ? Do two wrongs somehow make a right !

02:13:40 (Co-Pilot) Climb… climb… climb… climb… 
02:13:42 (Captain) No, no, no… Don’t climb… no, no.
02:13:43 (Co-Pilot) Descend, then
(Whilst all the while the other co-pilot has his stick pulled back anyway ?!?) 

Fly-by-wire is great until the pilots are unaware the various overrides – that prevent them doing stupid things – have been switched-off.

[T]he crash raises the disturbing possibility that aviation may … be plagued by a subtler menace, one that ironically springs from the never-ending quest to make flying safer. Over the decades, airliners have been built with increasingly automated flight-control functions. These have the potential to remove a great deal of uncertainty and danger from aviation. But they also remove important information from the attention of the flight crew. While the airplane’s avionics track crucial parameters such as location, speed, and heading, the human beings can pay attention to something else. But when trouble suddenly springs up and the computer decides that it can no longer cope�”on a dark night, perhaps, in turbulence, far from land�”the humans might find themselves with a very incomplete notion of what’s going on. They’ll wonder: What instruments are reliable, and which can’t be trusted? What’s the most pressing threat? What’s going on? Unfortunately, the vast majority of pilots will have little experience in finding the answers.

Very, very scary.

Been re-reading a number of “anthropic” pieces – prompted subconsciously I suspect by the death of Christopher Hitchens raising the god vs rationality debate in my mind again. And looking back at the Paxman / Hitchens interview from a year ago; Paxman suggesting that Hitchens own anti-theistic atheism is so dogmatic, it too is (was, sadly) a religion.

The seminal work on anthropic factors in science is Brandon Carter 1973. This quote sums up the dogmatic angle:

Copernicus taught us a very sound lesson that we must not assume gratuitously that we occupy a privileged central position in the Universe. Unfortunately there has been a strong (not always subconscious) tendency to extend this to a most questionable dogma to the effect that our situation cannot be privileged in any sense.

There ARE things weird about our position in the Universe that deserve explanation, rather than just being blown-off as self-centred anthropic delusions. See the Larry Kraus quote here. The reason to extract the Carter quote above is so I can throw it back at anti-anthropic arguments of that kind.

@GeorgeGalloway Contrasting two posts today.

This morning George Galloway tweeted his cynical disbelief that the execution of a Saudi woman yesterday for “sorcery” had gone unreported in mainstream (BBC) news. (He was right, it was easy to find reports elsewhere, even CNN, but it didn’t show up in BBC news searches.) Again whatever the morality of the death penalty, and notwithstanding it appears her “sorcery” case may have been genuinely criminal fraud of some kind – I couldn’t help thinking the response @GeorgeGalloway was – because with 75 such Saudi executions per year – one was hardly newsworthy.

Contrast that case for example with this one (from the BBC) later today:

At the [coal] industry’s height in the 1920s, 1.2m men were employed in the pits and approximately 2,000 died in accidents every year. Today there are only about 3,500 miners but each time there is a death it is national news. [There were “just” 17 UK pit deaths in the last 10 years.]

Never been a fan of management metrics – it’s all too easy for lazy managers to measure what is easy to count, and not deal with what really matters. Even Einstein said “not all that counts can be counted”. However this Forbes piece by James Slavet nails five valuable “metrics” worth assessing subjectively – starting with “Flow”.

[M]ost managers only measure outputs, not inputs, which is like telling a Little League team to score more runs, rather than actually explaining how to swing a bat and make contact with the ball.

BTW I couldn’t believe that the “you can’t manage what you can’t measure” adage could have originated with Peter Drucker, at least not outside some particular context. Good to see I’m not alone in seeing that message as alien to Drucker’s style. I certainly couldn’t find it as an attributable quote – sounds more like an adage created to justify Taylorism to me.

Good to see the actual quote (in a comment response):

“What you measure is what you get.”

As Ed says, “that is very different from the attributed quote. In fact, it is right on target. If you measure billable hours, you will get more billable hours”. You treat people like children / monkeys, you get children / monkeys. (I recall my Master’s thesis concluding something about the need for multiple objective and subjective measures in order not to skew behaviour towards narrow measures.)

Also like this from another actual Drucker quote:

Reports and procedures should be
the tool of the man who fills them out.

A particular bug-bear of mine is reports like time-sheet and expense report systems with UI’s that are formatted the way the report user wants to see things, not the way that is useful to the reporter – eg allocation to cost-centres or breakdown-codes before recording as line-items is a common fault designed to make form-filling twice the chore it needs to be. Reports should always be – this is what I did from my perspective as I report it – with the tools automating the re-presentation of the same information in whatever format management requires.

I only “rediscovered” Drucker relatively recently, since his death in 2006, and largely because I was taken by his debt to Mary Parker-Follett.