From: Strata Rose Chalup (strata@virtual.net)
Date: Fri Dec 22 2000 - 11:44:50 PST
Thanks, Jeffrey.
I think the fundamental problem facing the "P2P industry" is that there
really is no such thing. There's P2P technology, but no "industry".
How can that be, with all these players out there? Well...
Let's take the acronym P2P and expand it. Peer to peer, right? That
would be peer to peer WHAT, exactly? File sharing? Messaging?
Directory? Discovery? "All that and more!" trumpet various marketing
slogans, somewhat missing the point.
You can have a P2P technology company that produces a platform, a
library, an API, a protocol, whatever, for enabling P2P transactions.
Those transactions are going to be of various types and forms. They're
going to have different needs. Theoretically, you could say that
everything passing between two P2P nodes is a "message" and that a rich
messaging fabric with support for variable QoS is a P2P protocol. File
sharing, lookups, instant messages would all be instances of that.
Or not. Or you could set out to define a protocol for meta-sharing,
where you are passing meta-information and the P2P nodes are acting on
it, supporting arbitrary existing protocols. Forgive my utterly
unpracticed XML, probably full of format/usage errors:
<p2p-trans>
<originator>myhost</originator>
<method>directed-broadcast</method>
<interested-hosts>host1,host2,hostXYZ</interested-hosts>
<transtype>search</transtype>
<transopts>advanced, infix</transopts>
<payload>danger+kitty NEAR (rocket NEAR love)</payload>
<maxreturn>4000 bytes</maxreturn>
<returnmeth>HTTP,HTTPS</returnmeth>
<returnformats>HTML,MIME,TXT,XML</returnformats>
<authinfo>anonymous,SHA{null}</authinfo>
</p2p-trans>
Or maybe a P2P application is really a dynamic ad-hoc directory finder,
and then uses that info to then trigger protocols such as Napster,
Gnutella, Jabber, Tooltalk (couldn't resist!) and so on that are worth
supporting on a variety of platforms. So P2P applications might be a
combination of directory-reachers and proxy-facilitators.
BUT!
Here's the problem-- in none of these approaches is there any revenue
model that is sufficiently robust to both fire and sustain startup
development of an entire industry. If P2P applications are merely
discovery and peering of existing protocols, the IMPP phenomenon makes
it against companies best interests to cooperate to hammer out a
protocol. If P2P applications are a giant message passing bus, even
more so.
All that a P2P working group can do is essentially repliate the CORBA
process, and we all know how long THAT took, and most of why-- all the
major vendors wanted things that they already had implemented to be part
of the standard, and the things their competitors already had
implemented to NOT be part of it. And we've seen things like SOAP and
XML-RPC arise because of the sheer massiveness and unwieldiness of
CORBA.
Indeed, the "definition" of P2P is both elusive and somewhat pointless.
Everyone wants it defined enough so that they can say they "support" it,
but essentially P2P is like "messaging". Do you "support" messaging?
Well, umm, what KIND of messaging? AIM? ICQ? SMTP? POP? IMAP?
I believe that the P2P world has to fragment before it can come
together. Making existing protocols reachable by independent nodes via
resource discovery is a potential definition of P2P, but it's not one
that makes MONEY for anybody!
How do you make money? You sell software to access a custom protocol--
doesn't work if the protocol is open or if you don't own it. You
deliver additional content (ads/etc) with the results of a query--
doesn't work if the content payload is pure peer-driven. You salt
results transparently and preferentially to content sources that pay
you-- doesn't work if query sites are pure peer-driven.
If you don't have a central server or a custom protocol, where are you
going to make your money? Selling development kits to folks who haven't
figured that out yet? I don't think so. You can have a steady business
selling development kits to folks who *have an existing peer community*
that they wish to enable-- folks like big automakers, telecom giants,
retailing giants (Sears P2P net anyone? YardSaleAmerica.com? I made that
up, but it would work..), and the like.
Well, I hear you say, those are FINE customers. We can make money at
that!
And so you can, and I don't at all wish to discourage anyone from
trying. It's a great market segment that can sustain a company very
profitably for many years.
BUT!
I will point out that the whole reason there's a P2P "phenomenon" and a
P2P working group and all that, is that folks still seem to think that
there is a goldrush opportunity to collect *individual user fees from
individual users* in connection with P2P.
I think that is a highly unproven assertion, and that the particular
opportunity doesn't exist. I think you can sell software licenses for a
few million users at .10 - .50/seat, depending on value-add, on an
annual basis, which is a very nice chunk-o-change. I don't think anyone
is going to go out there with a P2P "killer app" and collect even $5/pp
from each of millions of users.
And the latter hope/expectation, I believe, is what's fueling the P2P
hype. Just ain't gonna happen.
"What about Napster? What about the possible Bertelsmann deal? All
those Napster users bringing in $3.95/month each! Slaver, slaver,
drool!"
Sorry, that case is morphologically equivalent to "I am a
cable/AOL/Earthlink subscriber, I am paying for access to a premium
channel". In order to get at those juicy per/head money numbers,
somebody had to spend a LOT of money building an ISP/service
infrastructure to support those users. In any possible structuring of
that deal, Napster will get a fixed amount per user (or user tier of N
users) from Bertelsmann, and still be on the hook for all the
infrastructure, support, and development costs of running a big ISP.
Especially the SUPPORT costs, which are a killer in the ISP industry.
Once folks are actually paying for something, you must support it, and
that means a whole level of help desk/800 number/escalation path stuff
that Napster does not have now and will have to build or outsource
greenfields, at a cost probably equal to 1.5 - 2x their entire physical
infrastructure cost. Since helpdesk type support does not, sadly, scale
onto cheap hardware the same way that Napster does. Despite some firms'
attempts to try to do so. :-) If the Bertelsmann/Napster thing goes
through, Bertelsmann will be water-sliding through fountains of cash,
and Napster's troubles are just beginning.
So the dirty little secret of P2P is that all the really AWESOME revenue
models depend on some poor sucker out there running the equivalent of an
ISP and soaking up all the infrastructure and support costs of running
the centralized or pseudo-centralized service. There's enough open
protocols and open-source FREE P2P stuff out there that you can't do a
closed-protocol user client play to collect your per-user dollars.
That's called a "browser war", kiddies, and we've already seen that Not
Work, Spectacularly.
Real Networks might have missed a great opportunity-- if they'd
partnered with Napster (don't know if that was actually possible given
timelines) or invented Napster, they could have had a P2P "hit" where
folks would buy RealPlayer to access the sites streaming the audio.
Wups! But who knows, it might have already been too late for that kind
of a ploy.
Actually, in principle, the "browser war" does work, a little bit, but
for existing players only. So AOL, MSN, Sprint, and the like, who have
proprietary branded services with huge subscriber bases already, might
build an in-service better P2P mousetrap and have signups beat a path to
their door. Their incremental cost over their EXISTING
infrastructure/support will be miniscule, and they'll make money nicely
with new subscribers. That just won't work for the folks who aren't
ALREADY that huge. The pure-portal players could be on either side of
the line, hard to tell-- of course, most of them are ad-revenue or
cobranding or clickthrough commission based, so adding a new service
doesn't help them unless revenue comes with it. It's certainly sure
that all the new, small firms, or "free now but someday we'll put up
really reliable servers and folks will pay to stay on" firms, those are
fully Out of the Game in this.
So-- the bottom line? P2P is alive and well. P2P is going to be
exciting, huge, and thriving, and I'm really interested and excited to
see what new things are going to come out of this. Ad-hoc friends and
family networks. Dynamic VPN multicast structures where you can share
all kinds of things with your chosen peer groups. Great stuff! P2P may
evolve to the point where I actually cave in and get an AOL account to
use some damn thing that my non-tech family already uses, like I would
have with AIM if AOL hadn't bought Netscape and spared me the trouble.
:-)
P2P application toolkits, servers, and dedicated protocol-only
outsourcing services? Definitely gonna happen, gonna create some good
companies, gonna be a traditional software development and/or
fee-for-service outsourcing model. Some things will fly, some will drop
like a rock, some will dig halfway to the chewy molten center of the
mudball, claiming all the way that they're really climbing rather than
digging. Frameworks to let non-programmers start ad-hoc services
through browsers, frameworks to let perl and python and visualbasic
programmers build P2P apps that depend on existing service infrs,
big-ass server farms driven by completely unglamorous hardworking
protocols brokering and centralizing and delivering P2P payloads
transparently for somebody's customer base. More great stuff!
A new crop of P2P "browsers" that people pay a fee to use? P2P "services
revenues" that are collected by companies without a huge scary
all-consuming ISP support/infr-costs blackhole? Dead as a doornail, too
dumb to stop breathing. Sorry!
So, what if you're already in P2P and aren't satisfied with the chance
to build another Phone.com or Tibco? $$Millions$$, someday $billions$,
of dollars of sustainable revenue, obtained only after some
anachronistic hard work and passage of years rather than spun out of air
and dew and viral marketing? That just DOESN'T make your heart beat
faster? Oh well!! Leave your business plan under your pillow and maybe
the Sympathy Fairy will put a sawbuck under it before making off with
your stereo and DVD collection.
Of course, if anyone reading this is a carpet-bagger at heart and only
cares about getting in for a year or so and Personally Getting RICH,
hell, there's still a LOT of opportunities in P2P. I'm just engaging in
my usual self-limiting behavior, naively assuming that folks are
interested in building a sustainable business that will outlast their
involvement with the company. I'm just old-fashioned. Y'all know that
by now. :-)
Cheers,
_Strata
PS- Thanks for the ptr to the next P2PWG meeting. I don't suppose
they'd appreciate someone with my viewpoint getting involved, and naked
Emperors aren't usually very aesthetic. But if you think there's a real
chance that the hype will die down enough to define
payload-protocol-independent meta-transport mechanisms, I'd be very
interested in getting involved. There may not be piles of dollars
there, but that's where the future of a lot of net services could go, in
a very positive direction.
PPS- Think somebody else needs to see this, either to gasp in awe at the
wisdom contained, or to gasp in awe at the cluelessness displayed?
Forward away, with my blessings, just keep it intact.
-- ======================================================================== Strata Rose Chalup [KF6NBZ] strata"@"virtual.net VirtualNet Consulting http://www.virtual.net/ ** Project Management & Architecture for ISP/ASP Systems Integration ** =========================================================================
This archive was generated by hypermail 2b29 : Fri Dec 22 2000 - 12:37:51 PST