Saturday, 28 September 2013

the difference between superstition and science

Science generally makes statements or theories predicated on evidence. Most of the time, scientists won't claim that they have a fact, but rather a correlation. A correlation tells you that two things co-relate. That is, as one changes, the other changes. You might recall y = f(x) from mathematics at school; this means that the y axis is a function of the x axis, or correlates or changes with the x axis.

Many people misunderstand scientific findings and think that scientists are often giving facts (causal relationships between x and y, say), when in fact they're just giving correlations. Consider booze, violence and poverty. In such a case, if there's a correlation, it's not an established fact that booze causes violence when coupled with poverty. Why? Because counterexample anecdotes can be found that fall outside of the booze - is - proportional - to - poverty - and - violence graph; these are called 'outliers'. Thus the theory is only generally accurate rather than a fact. Consider a rich guy who beats his wife. Booze in this case might not be involved, nor poverty. Therefore booze and poverty are not necessary or sufficient causes of assault. 

A cause is a correlation for which a mechanism can be found. A correlation is a relationship between two sets of data for which no mechanism has yet been found. The job of scientific theory is to supply that mechanism as a theory.

Now: What is superstition? I venture the following definition. A superstition is a belief based on either (a) anecdotal evidence, i.e. too few cases, or, (b) a belief based on a false assumption that causation exists where there's merely anecdotal correlation.

So, here's an example of a superstitious remark: My aunt died when a black cat crossed her path, therefore, black cats are to be avoided because they are harbingers of death. No. That's just one anecdote; so the correlation between death and black cats is anecdotal, and the theory that derives from it, is a superstition because (a) it's not giving a mechanism, and (b) it's one anecdote; we need a few thousand. 

The key differences between a scientific correlation (a law or theory), and a superstition, is the presence of a causal mechanism and quantifiers. So the following description is NOT superstitious: My aunt believed in the black cat superstition. She was driving one day, and a black cat crossed her path. Due to her panic at the sight of the black cat, she swerved into a concrete pillar at sufficient speed and mass to cause sufficient impact velocity, and died. So, in a sense, the black cat could be said to have been the cause of her death, but actually her fear was the cause of her death; the sighting of the cat was merely one component. Had she been a cat lover, say, and had a white cat had crossed her path, she still would have swerved, and she still would have died. Therefore, it is not the black cat that was the issue, but any cat at all, of any colour.

See the difference? One story connects specifically a black cat with specifically one death, and claims that that means that a general rule about black cats is true. That's also called 'fallacy of affirming the consequent': You see a black cat. When you see a black cat, you will die. You died. Therefore it was because of the black cat. In the scientific case, we have the same logical structure (a fallacy or generalisation), but look at the difference! Here we go: The scientific story connnects ANY cat with this specific death by explaining in terms of the car's momentum and speed, and the driver's mortal fear of colliding with the cat. She swerved, therefore, given her momentum, speed and direction, she hit a pillar and died. This model quantifies (gives speeds, momentum, etc), and it explains the mechanism.

Monday, 23 September 2013

ban on booze adverts - a worthwhile experiment

As with the stupid argument that "scientists should stop worrying about getting to Mars and start worrying about cancer" - which misunderstands that different scientists do different studies.... this argument, that "the government" should "worry about more important crimes" fails to understand (a) that we're in a federal system, so the provinces do whatever they like, and (b) that the health ministry has nothing to do with department of justice and constitutional development. That's like saying, "Oh, the minister of health should stop worrying about booze ads and start worrying about the matric pass rate".

The only connection between DoH and DoJ is that most violent crimes apart from hijacking and robbery are booze-induced, e.g. murder and spousal abuse and rape. If these crime stats go down then the Government WOULD be doing something about more serious crimes. 

The fact that the minister is trying to ban alcohol adverts suggests that the minister is far from corrupt, since no amount of 'incentives' from the alcohol industry has prevented this action.

The mandate of the department of health is to ensure south africans don't get unhealthy, and what to do when they do. Preventing booze advertising is just one example of a step or attempt to do so. If the ad ban fails, ie alcohol abuse does not decrease, then the ad ban is not justified. If however alcohol abuse does decrease after the ban, then the ban worked. The only way to find out is to enact the ban and see what effect it has, if any.

People talk about job losses: well, you know who will lose their jobs? The 30 or so rich people who place adverts for alcohol companies who have the hundreds-of-millions-of-rands "account" to manage placing adverts for these companies. The ad industry will lose some revenue. Big deal. They're all rich white-collar workers. This is a pro-poor action to discourage poor people from seeing booze as glamorous. Have you not seen that ridiculous advert for whiskey which says "one day this will be you" or words to that effect with a Learjet??

php 7 nightmare

OK so Centos 6 insists on installing php 5.3 and even if you download other RPMs and install them, they do not replace the existing 5.3 whic...