Frequently-asked questions
Why are you charging for something
that used to be free?
Rember, the only thing that costs money is the complete
documentation for NumPy. This book never existed before
(documentation can be a problem even for good free software). For
most people, it is difficult to get motivated to use spare time to
write documentation (which can be a tedious task). Spare
time, by definition, is the time you don't have to spend earning money
to provide the necessities. Most of us would rather
use spare time to do something fun. Rather than not have
documentation, the idea of charging for it (so that non-spare time
could be spent on it) was born.
Free is great. Who wouldn't turn down something good that's
free. However, because of the fundamental problem of scarcity,
everything can't be free. In other words, it takes effort to
produce things.
Money has emerged over millenia as a means to organize the behavior of
billions so that
other people work to produce the things that we all want. If you
pay for something, you are telling the network of workers that you are serious about
your preferences. Prices emerge from this system of
voluntary transactions based on how much people actually value
something relative to the other things they value.
Without these price signals (i.e. if
everything were free), then the problem of scarcity would still remain,
but now nobody would know what other people honestly value. We
would all have to just guess. As a
result, regardless of our best intentions, it would be easy to end up
spending all kinds of time doing things
few people actually want or need. Functioning voluntary
markets are essential for prosperity.
So there is truth to the old adage of "you get what you pay for" in
that your money serves as a signal to everybody in the world that this
is what you really want people to spend more time doing.
For example, if the NumPy documentation has little value to people then
that is an important signal that something must be wrong with
NumPy. Either developers need to spend time fixing the problem,
or
abandon all but perhaps a little spare time on the project.
Without a price signal, nobody really knows.
Will a later version of the NumPy
software stop being free?
No, No. Despite the
very good reasons for prices to exist in the world, using restricted
distribution for open source software won't work because it is then no
longer open source. Charging for the
documentation is not some
veiled attempt to close the
source code and start charging for it. There will
always be a free version of NumPy available as long as people are
using it. The code is open and completely free.
Anybody who wants to can download the code, learn all they desire about
how the system works by studying the source code and internal
documentation strings. Additionally, at least a dozen people
spread all over the world are involved in the maintenance of NumPy. It is highly
doubtful that any company would want to buy off all of those people in
order to convince them to stop working on a free version of NumPy.
Even if a company managed it there would always be others
who could take the code and continue.
Isn't your total price too high?
It's possible. Prices are always a matter of fluctuation --- the
old supply and demand notion. If the
price is too high, then the price actually agreed upon in the end will
be a signal to others regarding what the current value of NumPy
is. So, it's actually in the best interest of those using NumPy
for the price to be high. This will signal other entrepeneurs to
use their non-spare time to make NumPy better. Relying on
the volunteer labor of people will only get you so far (of course there
are other indirect ways to fund software development with non-spare
time as well).
Isn't your per-copy price too high?
Maybe. If you have a specific offer, email me and let me know
your offer. Again, time will tell. Prices are a
trial-and-error process. For some people the price will be too
high. Others will be willing to pay it in order to acquire
something of value. A price system lets those who value something
the most receive it.
Is your license saying that if I
buy the electronic copy only one person can ever look at it?
No, of course somebody can look over your shoulder at your copy.
You could even send it to them for a trial
read, or copy portions under
"fair-use." The point is that if somebody needs to use the
book regularly, then they should buy their own copy. You
should not just copy your book for them (electronically or otherwise).
Is there something like a
site-license so that everybody in our organization can access the
documentation?
If you buy 20 copies of the book then you can share it with anyone in your local
organization (site-license). You will also receive the html version when it is ready
that can be posted on an internal-only webserver that is not visible from
outside your institution (out of the glare of Google).
Honestly, though, if your institution is using Python for scientific
and technical computing, it is in your best interest to purchase as
many copies of the documentation as you can to encourage its further
development.
Will there ever be a bound book
that I can buy?
Yes. Details forthcoming. Those who have purchased the e-book will be able to
buy the bound book at cost.
Doesn't requiring payment for documentation "shoot-yourself in the foot."
From time to time, some complain that because the documentation is (temporarily) non-free then people won't download it and use it. First of all, there is all kinds of "free" documentation. Just as much or more than Numeric in fact. Look at http://www.scipy.org for examples and in the source itself (there is a doc directory). Second of all, the book goes beyond "documentation" and so is better considered as a book describing NumPy. People may continue to use Numeric and Numarray but the developers are not. It will be harder and harder to get help as the standard answer will be "use NumPy." Time will tell, of course.