Re: is free, open source software ethical?

In article <I_KdnWAVIuXHKVLanZ2dnUVZ8q6onZ2d@[EMAIL PROTECTED]
>,
 Richard Heathfield <rjh@[EMAIL PROTECTED]
> wrote:

> Ron Garret said:
> 
> > In article <aJCdnaEZBNjRElLanZ2dnUVZ8s-qnZ2d@[EMAIL PROTECTED]
>,
> >  Richard Heathfield <rjh@[EMAIL PROTECTED]
> wrote:
> > 
> >> > Random data in fact cannot be compressed.  That is actually part of
> >> > the definition of random data.
> >> 
> >> This is equivalent to claiming that, if data can be compressed, it
> >> cannot be random.
> > 
> > Randomness, like friendship, is an continuum, not a dichotomy.
> 
> What do you mean by this? Do you mean that the randomness of data is 
> measured by how much it can be compressed?

Yes.

> >> If you are right, we can sometimes predict with 100% accuracy
> >> the next value in a random stream of data, such as - say - the toss
of a
> >> fair coin.
> > 
> > Huh?  How do you figure that?
> 
> It's a sort of reductio ad absurdum. Let's start by assuming that random

> data *cannot* be compressed. That's your claim, right?

Yes.  By definition.

> Now, either there is such a thing as a process for generating random
bits, 
> or there isn't. If it is your further claim that there is no such
process, 
> then you are effectively arguing that there is no such thing as 
> randomness.

No.  It is possible to prove that incompressible data exists.  But the 
proof is non-constructive.

> So let's agree that we have a process for generating a genuinely random 
> stream of bits

OK, but for the record this is a highly dubious assumption.

> We examine the first N results, and then try to compress { N results + 1

> head }. Now, either we can or we can't.

By definition, we can't.

> If we can't, well, we can't.

And we can't.  So I guess we're done.

> But if we *can* do so,

But we can't.

> then by your argument the data cannot be random if the 
> next result is a head, and yet we are using a genuinely random stream of

> bits, so the data must be random, and therefore the next result *must*
be 
> a tail - which means that we've predicted the next bit with guaranteed 
> success, which means that the stream can't be random - and this 
> contradicts our initial assumption.

Well, it would if we could, but we can't so it doesn't.

> Therefore, either the stream isn't random after all (and this argument 
> applies to *any* method of generating random bits), or the assumption
that 
> random data cannot be compressed is false.

No.

> >> Let's say for the sake of argument that we can compress 10 or
> >> more identical results (10 heads or 10 tails).
> > 
> > If you want to take issue with the claim that random data cannot be
> > compressed you seriously undermine your case if you begin by assuming
> > for the sake of argument that it can.
> 
> It is not part of my argument that 10 (or whatever) heads is or is not 
> random. My argument uses the fact that it is not only possible but
trivial 
> to compress repeated data if there are enough repeats.

Really?  Describe to me this "trivial" process.  And also tell me 
exactly how many repeats are "enough".

> Since you claim 
> that random data cannot be compressed, you are arguing that random data 
> cannot contain sufficiently many repeats to make compression possible.
My 
> point is that, if that is true, we can use the fact to make occasional 
> predictions about random data with 100% success. That is, if we're
allowed 
> to *choose* when to make the bet about what the next bit will be, we can

> guarantee that we will win that bet! And that's absurd. So one of our 
> assumptions must be wrong - either there's no such thing as random data,

> or random data can occasionally be compressed.

Of course random data can "occasionally" be compressed.  But whatever 
scheme you use to compress it "occasionally" will also "occasionally" 
make the data *longer* if you encode it using the same scheme.  In the 
long run, the net result of any scheme you choose will be no net 
compression if the data are random.

> Obviously the fewer the repeats, the harder it is, and the less likely
that 
> it can be done. (10 is very low, and was merely a frinstance.) But
there's 
> nothing magic about 10 in my argument.
> 
> I generated a run of 1600 heads (where 'set bit' = 'head'), and shoved
them 
> through zip, which reduced it to 1216 bits (including all the file
header 
> information and stuff - the actual compression reported by zip was 97%;
it 
> was the overhead that made the end result so large).

You were probably using a pseudo-random number generator, and what you 
are observing is (almost certainly) the difference between pseudo-random 
and truly random.  Generating truly random data is quite difficult.

Here's a little though experiment for you: what would happen if you ran 
the output of gzip back through gzip again?  Would you get even more 
compression?  What if you repeated the process a third time?  Is there 
ever a point at which you would no longer get any compression?

> >> > To say that the sun revolves around the earth is very
> >> > badly wrong in fundamental ways.
> >> 
> >> Experimental evidence suggests very, very strongly that the Sun goes
> >> round the Earth. I seem to remember that you're very hot on
experimental
> >> evidence.
> > 
> > You have a deep misunderstanding of the nature of experimental
evidence.
> > It is true that (some) experimental evidence suggests that the sun
> > revolves around the earth.
> 
> In fact, for the vast majority of people it is the only evidence they
have.

True.  That is most unfortunate.  But the mere existence of a lot of 
ignorant people does not change the underlying truth.

> > But there is other evidence that is, at
> > least as far as anyone can tell, *incompatible* with the theory that
the
> > sun revolves around the earth (e.g. the seasons, the earth's
equatorial
> > bulge, the retrograde motion of Mars, etc. etc. etc.)  So the fact
that
> > there is evidence that suggests that the sun revolves around the earth
> > is *irrelevant*.  Such evidence does indeed exist.  Nonetheless, the
sun
> > does not revolve around the earth.
> 
> I'll alert the media. But in fact you have completely missed the point, 
> which in this case is about frames of reference. To an observer on the 
> earth (and that is where 99.eversomany9s% of all known observers have
ever 
> been), it is the Earth that stands more or less still - relative to the 
> observer! - and the Sun that moves. The other evidence you mention does 
> not mean that the Sun does not go around the Earth - it merely means
that 
> looking at it in that way might not be the most helpful model. And
they'd 
> be right. In fact, I am very, very sure that they are right.

Until you want to understand things like the seasons.  Then you're in a 
real pickle with a geocentric model.

> Do you believe in centrifugal force? I would guess not. And yet, if one 
> chooses one's frame of reference carefully, it /does/ exist.

I don't even know what it means to "believe in" centrifugal force.  
Centrifugal force is a theory that has a certain amount of explanatory 
power, but not as much as other competing theories.  Does heat exist?

> >> > It is likely
> >> > that your good deed strengthened your friendship in ways that will
> >> > provide you some direct benefit down the road.
> >> 
> >> ...but likely false. I had all but forgotten the incident until this
> >> discussion came up about free software, and I'm sure my friend has
> >> forgotten it too by now.
> > 
> > Maybe.  Or maybe some incident will occur that will trigger this
memory
> > in your friend just as this incident has with you.  Maybe some day
> > you'll be in a bind and your friend will suddenly remember that back
in
> > the day you did her a favor and that will push her over the edge to do
> > you one in return.
> 
> It's possible, but (a) it's unlikely, because neither she nor I think in

> terms of transactions where friendship is involved, and (b) it's 
> irrelevant, since it didn't form part of my reasoning when I agreed to 
> help her. I agreed to help her because she needed help, not because I 
> expected to get anything out of it.

OK... I need help.  I need $100.  Will you give it to me?

> >> In any case, the objection is simply wrong. Altruism is about intent.
I
> >> will gladly concede that some free software developers give away
their
> >> code in the hope of some gain. But there are certainly those who give
> >> away their code because they want to benefit others. That *is*
altruism
> >> in action.
> > 
> > Can you cite an example?
> 
> I did already.

I must have missed that.  Who was it?

> > Certainly RMS is not such a person.
> 
> Whether he is or whether he isn't, the existence of people who write 
> software to give away for motives other than altruism does not imply the

> non-existence of people who write software to give away altruistically.

True.  I was just hazarding a guess as to whom you might cite.  I'm 
still waiting to learn who these altruistic coders are.

rg