[LAD] RAUL?

[Fons Adriaensen]

On Wed, Nov 16, 2011 at 01:40:36PM +0300, Louigi Verona wrote:
> "The reason for turning something into proprietary
> is money."
> 
> Not necessarily. To be more general - the reason to turn something into
> proprietary is to have control
> over people who use your application. Whether you want to use that control
> to extract money or do something
> else is your choice.

The final goal of a commercial firm is to make money. We are
not talking culture here but business. And for me, to make
money by selling something other people need and are prepared
to pay for is perfectly OK as long as the game is played to
fair rules. Which means you pay the people who contributed
to creating whatever you sell. I'm pretty sure that the bloke
at MikrosImage who contacted me is not working for free, yet
he expects me to do exactly that.

The code in this case was the set of algorithms used in Ebumeter.
Let's have a look at what it takes to write this.

* Obtain, study and fully understand the relevant ITU and EBU
  documents.

* Work out a set of algorithms that implement them in a practical
  way (which means some minimal requirements on efficiency, both
  CPU and memory, and portability).

* Since this is something that could be used in many different
  applications, create an API that facilitates this.

* Write and debug the code.

* Test performance against the requirements stated in the
  standard documents.

* Document the implementation and test suite.


As a SW development manager working in a commercial environment,
and given a competent programmer who also knows some audio DSP
basics, I'd allow a man-month. And when I was actually doing that
job I was often said to be too optimistic in my estimates.

That adds up to something like 3 to 4 kEuro. Most manageers, if
they can instead buy the same for 300 Euro would do that. Unless
they know they can get away with stealing it. 

Ciao,

-- 
FA

Vor uns liegt ein weites Tal, die Sonne scheint - ein Glitzerstrahl.