Re: Languages don't matter. A mathematical refutation

["Jean François Martinez" <darkquark99@gmail.com>]

On Thursday, March 26, 2015 at 2:43:02 PM UTC+1, Maciej Sobczak wrote:

> At some point in time programming was introduced as a subject in the schools that precede university. You know, just as today some people are trying to push programming to nursery schools. The level (and therefore age) when programming is first introduced becomes lower and lower with time and this means that at the university level, as years go by, you can expect students to have higher and higher initial programming experience. Years ago a typical student entered university with zero programming experience and now this initial experience is significantly higher, thus allowing students to be more successful with exercises if the exercises do not change. And since such changes in the scholar system are introduced in a step-wise manner, you can also expect that at some particular time in the past the batch of new students had a step-wise more experience at the entry.
> 

Good objection.  It would be valid between, say, people having entered university in 1985 and people entering in 2015. But from one year to another success rate went from 0 (zero) to 50%.  Do you _really_ think in one year of programming in Visual Basic in high school, that's s one year where they also had to attend to a number of disciplines unrelated to programming, made such a difference?

> Unless you prove that this had no influence on the results in question, I refute the whole of your mathematical proof.
> 

See above.

> > This demonstration was just for fun
> 
> Sure. Preaching to the choir is always fun. ;-)
> 

No, it isn't. More than once I have battled with a particularly stubborn macho programmer.  I have already tried to use (another kind of) math on him to no avail.  I wrote that paper as propaganda material then I thought about him and knew he would just shrug it off.  "There is no hollow charge shell able to prevail against brains made of cast iron".  Solsyenitsyn. The smilie and the fun part was to hide despair. 



> But more seriously, it would be interesting to flip languages every year and see whether the advantages of Ada hold in terms of better results. Or run two classes in parallel (with random assignments of students to each class) and compare results over the years.
> 

I know.  But it is not possible.  We have to take what we have.  Also one thing I haven't discused is the "parapsychology effect".  One proponent of
parapsychology carefully collected examples who were too good to be just random
noise.  He was honest and very scientific so people took him seriously.
Problem is: the people who were gfeeding him data sent him only the experiments in which the subject had got high scores.   In fact it was the problem of getting heads twenty consecutive times.  If you play 2**20 games then you have over 60% chances of getting it.  And if you play 10 million games the chance you have of _not_ having 20 consecutive heads is inferior to 1/20000

But we don't have had thousands and thousands of people who had groups of students using C and Ada.   So when someone  reports a diffeernce who has a probability inferior to 1 in 10**5 it cannot be dismissed as "parapychology efect".  


One point however is that I need to do some fact collecting.  Like if university curriculim made that students of Professor McCormick had previous experience in Ada and not in C.  

BTW:  For pre-university experience.  In late eighties there were cheap C compilers like Borland's so a motiveted high school teenager would have
been able to afford.  Even if Gnat had been availbale it is far more dubious he would have used it to program in Ada.

> -- 
> Maciej Sobczak * http://www.msobczak.com * http://www.inspirel.com