Topic: is Greg Maxwell wrong about the block increase? (Read 4413 times)

On Sun, Aug 30, 2015 at 1:43 AM, Peter R  wrote:
> Dear Greg,
>
> I am moving our conversation into public as I've recently learned that
> you've been forwarding our private email conversation verbatim without
> my permission [I received permission from dpinna to share the email
> below that proves this fact].

Unfortunately, your work extensive as it was made at least two
non-disclosed or poorly-disclosed simplifying assumptions and a significant
system understanding error which, I believe, undermined it completely.

In short these were:

* You assume miners do not have the ability to change their level
centralization.

 -- In fact they do, not just in theory but in pratice have responded
    to orphaning this way in the past; and it is one of the major
    concerns in this space.

* You assume the supply of bitcoin is infinite (that subsidy never
declines)

 -- It declines geometrically, and must if the 21m limit is to be upheld.
    (Though I think this is not equally important as the other concerns)

* You argue, incorrectly, that amount of information which must be
transmitted at the moment a block is discovered is proportional to the
block's size.

 -- Instead the same information can be transmitted _in advance_, as
 has been previously proposed, and various techniques can make doing
 so arbitrarily efficient.

[I would encourage anyone who is interested to read the posted off-list
discussion]

I contacted you in private as a courtesy in the hope that it would be
a more productive pathway to improving our collective understanding; as well
as a courtesy to the readers of the list in consideration of traffic levels.

In one sense, this was a success: Our conversation concluded with you
enumerating a series of corrective actions that you would take:

--------
> 1.  I will make it more clear that the results of the paper hinge on
> the assumption that block solutions are propagated across channels,
> and that the quantity of pure information communicated per solution
> is proportional to the amount of information contained within the block.
>
> 2.  I will add a note [unless you ask me not to] something to the effect
> of "Greg Maxwell challenges the claim that the coding gain cannot
> be infinite…" followed by a summary of the scenario you described.
> I will reference "personal communication."  I will also run the note
> by you before I make the next revision public.
>
> 3.  I will point out that if the coding gain can be made spectacularly
> high, that the propagation impedance in my model will become very small,
> and that although a fee market may strictly exist in the asymptotic
> sense, such a fee market may not be relevant (the phenomena in the paper
> would be negligible compared to the dynamics from some other effect).
>
> 4. [UNRELATED] I also plan to address Dave Hudson's objections in my
> next revision (the "you don't orphan your own block" point).
>
> Lastly, thank you for the note about what might happen when fees >
> rewards.  I've have indeed been thinking about this.  I believe it is
> outside the scope of the present paper, although I am doing some work
> on the topic.  (Perhaps I'll add a bit more discussion on this topic
> to the present paper to get the reader thinking in this direction).
--------

To the best of my knowledge, you've taken none of these corrective
actions in the nearly month that has passed.  I certainly understand being
backlogged, but you've also continued to make public comments about your
work seemingly (to me) in contradiction with the above corrective actions.

Today I discovered that another author spent their time extending your
work and wasn't made aware of these points.  A result was that the other
author's work may require significant revisions.

I complained about this to you, again privately, and your (apparent)
response was to post to that thread stating that there was a
details-unspecified academic disagreement with me and "I look forward
to a white paper demonstrating otherwise!", in direct contradiction to
your remarks to me three weeks ago in private correspondence: In private
you said that your model may only hold in an asymptotic sense and that
"the phenomena in the paper (may) be negligible compared to the dynamics
from some other effect"; but in public you said /my/ position was
"academic".

At this point I thought continuing to withhold this information from
the other author was unreasonable and no longer justified by courtesy
to you, and I forwarded our complete discussion to the other author
with the comment "I'll forward you a set of private correspondence
that I've had with Peter R.  I would recommend you take the time to
read it and consider it.".

I apologize if in doing so I've breached your confidence, none of the
material we discussed was a sort that I would have ordinarily considered
private, and you had already talked about making the product of our
communication published as part of your corrective actions.
I do not think it would be reasonable to demand that I spent time
repeating the same discussion again with the other author, or deprive
them of your side of it and/or the corrective actions which you had
said you would take but have not yet taken.

(In fact, I would argue that you were already obligated to share at least the
substance of the discussion  with the other author, both as part of your
commitment to me as well it being necessiary for intellectual progress.)

As you say, 'we are all here trying to learn about this new amazing
thing called Bitcoin'; and I'm not embarrassed to error towards doing
that and aiding others in doing so, but at the same time I am sorry
to have done so in a way which caused you some injury.

In any case, your prior proposed corrective actions seemed sufficient to me.

It surprises me, greatly, that you are failing to see the extreme
practicality of what I described to you, and that it does not meaningfully
diminish miner control of transaction selection-- but even without it your
remark that the proportionality could be arbitrarily small (Rather than
0, as I argue) is an adequate correction to your work, in my view.

I believe my time would be better spent actually _implementing_ improved
relaying described (and describe what was implemented) than continue
a purely academic debate with you over the (IMO) inconsequential difference
between epsilon and zero.

Cheers,