Re: Learning Ada

[antispam@math.uni.wroc.pl]

Dmitry A. Kazakov <mailbox@dmitry-kazakov.de> wrote:
> On 02/10/2020 20:43, gautier...@hotmail.com wrote:
> > On Wednesday, September 16, 2020 at 1:09:57 PM UTC+2, Dmitry A. Kazakov wrote:
> > 
> >> I mostly agree with your analysis, except the last part. The problem is
> >> that the culture of programming and overall education became so low that
> >> it is no more a race against C++. C++ itself is in defense and losing
> >> against languages and practices so overwhelmingly bad that even C looks
> >> as a shining beacon. Winter is coming.
> > 
> > 20-30 years ago the pointy haired bosses would warmly recommend a buzzword called C++ (or something spelled "tchavasiblosblos" in some places) for A?ti projects, but now they all want a digital transformation strategy with blockchain and machine learning. I'm not sure if the level is really lower (we tend to make the past nicer in our minds). But clearly the notion of programming language was lost on the way :-) ...
> 
> Well, the method of ML used in these days goes back to an idea from 
> 50's. It was promptly discarded then as too weak to solve elementary 
> classification problems. The AI of late 80's was far more diversified, 
> advanced and powerful than this mockery, yet it died peacefully, because 
> of limitations it had. I don't even know what to think about all this.

Hmm, I would say that progress from perceptron in 50 to current ANN
is bigger than progress from FORTRAN to Ada.  Would you accept
critique of Ada based on experience with FORTRAN?  Anyway,
ML researches tried various methods and currently ANN give
best results.  Using methods from 80's it would be hard or
impossible to get results obtained by modern methods.

I would very much prefer more principled apprach than ANN,
but arguing against reality is not productive...

-- 
                              Waldek Hebisch