IP: A Wall Street take on the metered pricing thing

[Dave Farber <farber () cis upenn edu>]

From: "eunice" <eunice () ccgw Tudor Com>


I thought members of the list might be interested in the following.


Archives of Bill Gurley's newsletters can be found at www.upside.com


Eunice Johnson
Analyst
Tudor Investment Corp.
net: eunice () tudor com
vox: 617-772-4642
fax: 617-737-9280


______________________________ Forward Header
__________________________________
Subject: ATC 96-18
Author:  Bill Gurley <bgurley () abovethecrowd com> at ccgw
Date:    11/05/96 03:29 AM




Above the Crowd
Deutsche Morgan Grenfell Technology Group 
Newsletter Issue: 96-18
J.William Gurley
bill_gurley () dmgtech com
415-614-1159
     
USAGE BASED PRICING:NOT IF, BUT WHEN
     
"And you let go, let go, let go, 'cause you know you're getting 
tired. Can you feel it getting down to the wire?"
 - Buffalo Springfield
     
Regular ATC readers should know by now that we are not big  fans 
of the unlimited-use pricing model currently in existence on the 
Internet.  With each passing week, congestion problems  continue 
to  worsen.   Pacific  Bell (PAC, 34) now claims  that  at  peak 
usage,  one in six business calls do not get through as a result 
of  jammed  circuits  in the central office  switch.   Also,  an 
article last week on CNN Interactive declared, "Soaring Internet 
usage  is  bringing  the United States' phone system  perilously 
close  to  gridlock  by tying up millions of local  phone  lines 
every  evening."  Lastly, our own futile attempts to connect  to 
Netcom  (NETC,  15)  in Palo Alto, New York,  and  Austin  would 
suggest  that this ISP is running its systems right  up  against 
its limits.  We need a fix, and we need it now.
     
While  we  risk accusations of redundancy, we feel compelled  to 
reiterate  our  position  on  pricing.   We  have  a  hard  time 
understanding  why  so  many  people  fail  to  understand   the 
necessity  and  rationality of usage based  pricing.   When  you 
offer  unlimited  use  of  a  limited resource  overcrowding  is 
inevitable.    Of  course,  this  isn't  rocket  science,   it's 
Economics  101.  We think MCI's (MCIC, 30) Internet  guru,  Vint 
Cerf, sums it up best.  "The hill is overgrazed, there's no more 
grass, and the sheep die."
     
Once  you  accept  the wisdom of usage based pricing,  the  next 
question is "how do we get there?"  Intense competition  in  the 
consumer  Internet  market would seem to  preclude  a  shift  to 
clocked  services.   Nonetheless, we see  three  different  ways 
where  usage-based pricing could emerge without proactive action 
from your local ISP.
     
The first and most obvious way we move to usage based pricing is 
if  the  FCC  decides to levy a modest per-minute access  charge 
(perhaps  one-half cent per minute) on data service  connections 
from  the central office switch.  As many of you know, the RBOCs 
receive  a  hefty 3.25 cents per minute access charge from  long 
distance providers, but absolutely zilch from the ISPs.  This is 
despite  the  fact  that the average Internet  "call"  typically 
lasts  ten times longer than the average phone call.  A proposal 
from the FCC is expected in November with a ruling in May.  Once 
thought  to  be "out of the question," the FCC now appears  more 
amenable to a data service access charge.
     
If  the  FCC votes down Internet access charges, we will  likely 
achieve  data-voice price parity in an alternate way  -- through 
the  absolute  removal of access charges.  While  this  decision 
would assuredly be "free-market" friendly, it would also start a 
chain  reaction  that would likely result in per-minute  pricing 
for all telephone calls.
     
The  removal  or reduction of long-distance access charges  will 
unquestionably increase the demand for long-distance services as 
a  result  of significantly reduced long-distance pricing  (once 
again,  Econ  101).  This increased usage would  put  even  more 
pressure  on the already congested central office switch.   Once 
this happens, the RBOCs will be left with no alternative but  to 
move  to  per-minute pricing on all calls as a way  to  throttle 
demand.
     
Obviously,  a move such as this would be particularly  unpopular 
with  consumers and would re-ignite political concerns regarding 
universal  service.  However, keep in mind two  things.   First, 
most  countries  outside the United States do indeed  have  per- 
minute  pricing on local calls.  Secondly, when we designed  the 
U.S. telecommunications network, we made a decision to subsidize 
local   service  with  hefty  per-minute  long  distance  access 
charges.   Now, some want to remove the subsidy, but to continue 
to reap its benefits.  You can't have your cake and eat it too.
     
Even  if  the FCC fails to restructure access charges, we  could 
still  end  up  with  usage based pricing as  a  result  of  the 
continuing evolution of ISP peering relationships.  The Internet 
grew up in the altruistic world of academia, and the competitive 
effects  resulting  from commercialization  have  yet  to  fully 
materialize.   As  Hank Williams Jr. likes  to  point  out  "Old 
habits like these, are hard to break."
     
Most  ISPs  interconnect with each other at the  public  peering 
points,  also  known as MAEs (Metropolitan Area  Exchanges)  and 
NAPs  (Network  Access Points).  It is in these facilities  that 
ISPs  exchange data packets which are destined for each  other's 
network.   What  is  most  interesting is  that  the  "rules  of 
engagement"  for these public interconnect points are  nebulous, 
dissimilar, and quite likely to change.
     
Why  are  these peering relationships so likely to change?   For 
starters,  the  public  peering  points  are  one  of  the   key 
contributors  to  Internet congestion.  As with  any  "tree  and 
branch" architecture, the main trunk is likely to experience the 
most  traffic.  However, this is not the only problem.  Some  of 
the  larger ISPs are unhappy with smaller ISPs that interconnect 
at only a few of the public peering points.
     
Assume ISP A is a major provider with interconnections at all of 
the  major  peering  points.  Now let's  say  that  ISP  B  only 
connects at MAE East and MAE West.  If a customer of ISP B  asks 
for data from a customer of ISP A located in Chicago, ISP A will 
have  to carry those bits through San Francisco, even if  ISP  B 
has  a  presence in Chicago.  This is not only inefficient,  but 
ISP  A resents the fact that if ISP B were bigger (i.e. had more 
peering  points) that it would not have to carry these  bits  as 
far.
     
The  large  ISPs  have  already taken a significant  first  step 
toward solving these problems.  Through the use of what is known 
as  private  interconnect  points, these  vendors  have  created 
several 1-1 tunnels between each other's networks.  Some  people 
speculate that vendors such as MCI and AT&T (T, 35)    may  have 
as  many  as a hundred private interconnects between  their  two 
networks.   When  data  passes between these  two  vendors,  the 
public peering points never come into question.  Moreover,  data 
packets spend the majority of their time on the network of their 
own  service provider, as each ISP can off-load packets  to  the 
alternate ISP at the earliest possible point.
     
If  you have not heard about private interconnect points,  there 
is  likely  a  reason.  Out of fear of these arrangements  being 
considered  collusive by the public, the  FCC,  or  the  Justice 
department,  most of these arrangements have been secured  under 
NDA (non-disclosure agreement).   As such, it is hard to find an 
executive that is willing to talk about them.
     
Looking   forward,   we  think  the  ISP  world   could   change 
significantly.   Once a player such as MCI  can  create  private 
interconnect agreements with the five or ten largest ISPs,  they 
could  then abandon the public peering points altogether.   This 
would  obviously be a shocking move, but it would  significantly 
improve  the  performance of the MCI backbone.  Of course,  this 
would  isolate  several smaller ISPs, their customers,  and  any 
content  that was on those networks.  However, MCI could  simply 
state  that these ISPs are not carrying their fair share of  the 
backbone load and could encourage those ISPs to become customers 
of  MCI.  Once they became customers, they would then again have 
full  access to the network, albeit with MCI in control  of  the 
cost  of that access.  Of course, we use MCI only as an example, 
as any of the larger ISPs could push forward similar action.
     
Just  in  case you think this is all Oliver Stone-ish, we  would 
like  to provide you a quote from the prospectus of Digex (DIGX, 
10 7/8) a smaller ISP which recently went public.  "Although the 
company  currently meets those requirements (regarding peering), 
there  is  no  assurance that other national ISPs will  maintain 
peering relationships with the Company.  In addition, there  may 
develop  increasing  requirements  associated  with  maintaining 
peering with the major national ISPs with which the Company  may 
have to comply."  It is our impression that a major move such as 
the  one  speculated above would move us closer to  usage  based 
pricing.   In  such  a world, more ISPs would become  customers, 
allowing a few large players to dictate pricing.
     
If  you  think that is interesting, we would like to  leave  you 
with  one  more thing to consider.  We have heard from  multiple 
sources that MTV (VIA, 35) conveyed to Netcom that it would need 
to pay $1 per customer per year, if Netcom wanted to ensure that 
Netcom  customer's  could access the MTV  site.   While  we  can 
certainly  argue  over whether or not MTV's site  commands  that 
much  leverage, the idea of a content provider dictating  access 
pricing is surely something new to think about.
     
Above  the  Crowd is a bi-weekly publication focusing on the 
evolution  and economics of the Internet.  It is distributed 
through First Call, fax and email.   To be placed on the 
distribution list contact your Deutsche  Morgan Grenfell 
salesperson  or send email to atc-request () abovethecrowd com with 
the word "subscribe" in the body.  As always, feedback is both 
welcomed and encouraged.  ABOVE THE CROWD is a service mark of J. 
William Gurley.