[MD] Stacks

[Magnus Berg]

Andy and Krimel

On 2010-07-28 21:26, Krimel wrote:
> [Andy]
> I'm not sure how far this "stacks" thing will go, but I like it. I
> think it will prove a good static latching point, something to help us
> set aside contentious details and focus on the big picture.

One can always hope.

> Krimel on Levels in electronic computers:
> "I think the levels in the MoQ are a mess and Andy nailed them correctly
> earlier today with a comment something like we create levels on the fly. I
> have used that terminology several times in the past. Pirsig's claims that
> "level" are discrete and that they have conflict are just obviously false.
> Like all systems of levels they may have heuristic value, they kind of work
> as rules of thumb but they all break down when you try to put too much
> weight on them. Ten years of haggling ought to convince anyone that Pirsig's
> levels are particularly brittle in this respect."
>
> [Andy]
> Pirsig's MoQ stack is a very good stack. It makes a lot of sense. We
> can do a lot with it. (I used it last night to explain to people with
> no familiarity with MoQ my opinion on two different subjects: academia
> and law.) If we restricted our conversation to only that stack, we
> could have an unbounded number of conversations over many years.
>
> [Krimel]
> Since Pirsig lifts his stack from the organizational structure of most
> university curricula it is obviously very useful.

Don't know if you read my lashing out at Marsha over that word "useful" 
the other day. It's sooo defensive! You can make it mean whatever you 
want it to mean. So what you say is actually nothing. You only put as 
much faith in the levels as you want, and when it doesn't suit you 
anymore, you can just disregard it. That's "useless" in my book.

> [Andy]
> However, many more conversations are possible when you can package
> Pirsig's levels as a stack, set that aside, and go to work with other
> stacks.

Hmm... I sense here that when you say "other stacks", you actually mean 
other sets of levels, right? And each time you want to make a new stack 
for some particular purpose, you must reinvent the stack of levels and 
see what you end up with.?

I was rather hoping to be able to arrive at some "most general stack" 
that can be applied to any purpose. And then you can of course adapt 
that general stack for the context at hand to make it more concrete.

> [Krimel]
> Not to belabor the point but the creation of "stacks" is arbitrary, done
> with some specific aim in mind. Magnus's notion of a universal stack seems
> farfetched.

What I call the "universal stack" would not be this most general stack, 
but that general stack adapted to our universal physics.

> [Andy]
> The first step to understanding stacks in general is to look at what
> makes Pirsig's stack tick. Forget how he arrived at them and
> concentrate on what they compose together: a corpus of patterns
> arranged according to Pirsig's conception of evolution.

That's more or less what I tried to do in "The levels undressed". It's 
the universal stack and I tried very hard to see what makes it tick, 
i.e. the interface between each level.

> [Krimel]
> But Pirsig's claim that the levels are discrete and independent just doesn't
> fly. Despite Magnus's hope that they can somehow be made more discrete, I
> don't see that working out well. Look for example, at the OSI model in
> computer science. Even at the first layer of the stack, the physical layer
> there are blurred distinctions. What for example is BIOS. Is it software? It
> is hardcoded into a chip and while modern BIOSs are over writable the
> original BIOSs were not. And there is lots of "code" physically embedded
> within processors and other computer chipsets.

You have to come up with a better example to wreck my hopes. Of course 
BIOS is software. And microcode embedded in CPUs is also software. And 
as soon it's software, we can use the computer logic stack to analyse 
it. Of course it can make it easier to analyse the rest of the computer 
if we disregard microcode, or BIOS. But then we can simply change the 
purpose of the stack and limit ourselves to higher functions of the 
computer, but then of course we need to be really sure that the lower 
level functions work as specified.

A borderline I can think of is hardcoded, logic gate arrays where the 
output is not really computed by software but by lots of logic gates, 
but in that case we can simply not include that gate array in the stack, 
not in the software logic stack anyway.

Try to come up with a really good example from the universal stack. I 
can understand you doubting my big aims if you get stuck on a BIOS. But 
I applaud you taking up a new example. We see way too little of that 
here, only old replays from the books over and over again.

> [Andy]
> Can we just as easily set aside Pirsig's stack and each be the measure
> of our own circumstances? Try to work out another way to sort patterns
> into a different scale, then slice your scale into any number of
> levels---four is not a necessary number---or just leave it as a
> continuum. Then try to put your new stack (or continuum) to work.
>
> [Krimel]
> Right such divisions are created and sustained only to the degree that they
> are useful.

useful... Arghh! :)

> [Andy]
> Probably there are other ways to approach the production of stacks.
> This should be illustrated by examples from direct experience. I don't
> have anything to fill that space right now.

In my experience, it's not really that hard to find level borders since 
the rules of the levels have to be completely orthogonal, or mutually 
irrelevant you might say. For example, if you take the flip-flips that 
are made only with voltage levels and currents and produces rules such 
as and, or, not. There is no built-in rule somewhere that says that such 
and such a voltage should represent 1 and another voltage 0, such rules 
are made up by whoever is designing the level border, or in the case of 
where there is no designer, DQ, or simply chance.

It's the same thing with what I call the spatial level. Before it, 
before the big bang, there were no space, no time, no up, no down, left 
or right, no before or after or now. There were no mass because mass 
requires space to be in. But as soon as space itself was born, all the 
other stuff of the spatial level also appeared as a direct consequence. 
Mass *could* crash into eachother because first of all, mass existed, 
and 2nd, different masses could be at the same place at the same time. 
And gravity, what purpose would gravity have without space, without mass 
and without time? That's the type of discrete borders I'm talking about, 
not whether BIOS should be called software or hardware.

> [Krimel]
> Pirsig's example of the platypus comes to mind. Here was an exemplar that
> did not fit readily into the biology stack. Rather than throw out taxonomy
> on metaphysical grounds biologists just tinkered with the design of their
> stack and voila the platypus fit in nicely.

The thing is, if we chose our level borders carefully, and make the two 
levels orthogonal and mutually irrelevant, platypus kinds of unfitting 
things become impossible.

	Magnus