[Cryptography] Interesting discussion of Web 3.0 ...

Theodore Ts'o tytso at mit.edu
Thu Jan 13 23:42:11 EST 2022


On Thu, Jan 13, 2022 at 07:02:54PM -0600, Brad Klee wrote:
> 
> > It doesn't deal with computers as atoms because one of the core
> > arguments of the essay is that users don't want to run "servers".
> 
> This is what I call "one side of the debate". The other side is that people
> do want to run servers, especially to have fun doing "calculations" with
> their friends.

Sure, there are plenty of thigns that people will want to do for
"fun".  For example, there are people who are very happy using Large
Format Cameras, using real film, to take photographs, where you spend
15-30 minutes carefully setting up the image, and then taking the
black and white image, and carefully developing it in your own
darkroom.  However, the *vast* majority of people will just pull out
their smartphone, take a picture, and it will automatically be
uploaded to the cloud.

Will there people who want to run their own servers?  Sure, I'm one of
the weirdos that not only run my own web server, I even run my own
SMTP server.  But it's a massive pain in the *ss, and it's *not* my
primary e-mail address.  I don't even use it on the sending path,
because figuring out how to get other mailers to reliably accept
e-mail sent from my SMTP server as not being "spam" is takes too much
of my time.  I know how to do it, having once been one of the folks
responsible for postmaster at mit.edu, back when the MIT mailer was doing
more SMTP deliveries per day than AOL --- but the reality is I have
many more things to do with my time than to be a SMTP server admin on
a production basis for "fun".

These days, I just use my MIT mail address for most things, and MIT
has outsorced its mail system to a hosted Microsoft Exchange service.
MIT is no longer running its own mailers, and staffing its own
postmasters who used to compile their own sendmail binary, and
carefully kept it running and patching it as necessary.  It's much
cheaper to rely on a centralized service, and that is what the MIT
institution has chosen to do.

> 
> > It's really irrelevant whether you're talking about a $10,000 Intel XEON
> > server, or a $200 Intel NUC server; both are still... servers.
> 
> Did you mean "relevant" instead of "irrelevant"? Then I would agree with
> you. Not everyone is backed by an institutional super-budget, but many
> are curious what's out there in the computational multiverse.

No, I menant "irrelevant".  Whether it's a big computer or a small
computer doesn't *matter* from the perspective of being the system
administrator for the thing.

The *relevant* issue is that the vast majority of users aren't going
to want to run their own servers.  A few people will, sure.  But if
most people are going to be dependent on someone else's server
infrastructure, for *them* it's actually no better than just using a
pure centralized system.

The bottom line is that Cryptocurrencies like Bitcoin, and Web 3.0
distributed compuation, have a huge amount of overhead because they
can be distributed.  But if people don't actually get the benefits of
that distributed system, because it's much easier to download the
Coinbase Android or iPhone app, it's not clear the cost/benefit ratio
works out for those users who aren't running a full Bitcoin client and
directly participating on the blockchain on their high-end desktop
machine, and or running their own server which their phone app can
talk to.

> > But who is going to *run* these servers?  . . .  It's similar to . . .
> 
> Hopefully not one person. If the neighborhood is smart enough,
> there could eventually be numerous candidates up for election.

How well do most people know their neighbors?  Is someone going to be
trustworthy just because they happen to live next door?  OK, so
there's a small-scale server run by your next door neighbor, which is
running all of server needed for their neighbors' DeFi phone apps.

How do they know whether or not that person is *competent*?  How do
they know whether he will be dutifully installing all of the relevant
security updates?  How do they know that he won't abscond with most of
their life savings, leaving them to sob on the evening news, "I should
have stayed with Bank of America"?

Cheers,

					- Ted


More information about the cryptography mailing list