Again ... What is it going to COST?????

[bass at dmsd.UUCP]

		    What is stargate going to cost?

              A simple question with a difficult answer.


It's difficult for three reasons, all of which haven't been openly discussed
by the proponents of stargate -- but must have come to mind while tring to
figure out how to make a commercial service from usenet.

	1) As a commercial service stargate based news traffic WILL cost
	MORE money. The fee for most sites will be larger than the current
	costs to access news via uucp based usenet. This is a simple deduction
	since most sites currently recieve news free from another local site.

	2) As a commercial service stargate based news traffic CANNOT be
	rebroadcast to ANY OTHER machine or site without paying the fee for
	that site as well. This will be difficult to do since the current
	community ASSUMES free access and distribution. This is a simple
	deduction since the news link is planned to be scrambled AND without
	such a restriction the subscriber base will be less than a few
	hundred sites.

	3) The service is to be accessed via local cable operators who own
	the medium AND the vertical retrace time. They will want a cut of the
	fees at some point. This is a simple deduction since they have
	to make a living too -- and data transmission within the broadcast
	industry is becoming big business. IF they take 30-50% of the fee
	it WILL double the cost of service.

The current usenet is estimated at about 2,000 sites many of which are
multiple machines within the same organization -- I don't see most of these
sites paying multiple site fees for each machine. Particulary since the
largest segment of these sites are University, Government, and AT&T sites.

Of the current sites, many are operated by private individuals who access
usenet with a local tele call for FREE -- I don't see many of these folks
accessing stargate if the fee is $50/month, and almost none at $100/month.

To install and service 1,000 sites will require a staff of atleast 10 people
to handle sales, customer questions, cable operator questions, decoder shipping
, decoder repair/testing, and general admistrative functions. An estimated
fully burdened facility and staff budget probably exceeds $1,200,000 without
including fees paid for the satelite channel, cable operators, investors,
or other people with their hands out. At this level of service the low end cost
per user is about $1,200,000 / 1,000 * 2 (cable operator markup) = $2,400/yr.
I doubt that 50% of the installed USENET machines will become paying subscribers
in the first year.

Since customer service needs will grow with the userbase I don't expect
much economy of scale until year 2 or 3 where the installed base will become
a cash cow. The real costs are likely to be 3-4 times higher when you
include advertising, startup cash outlays, etc... it will take one big
subsidy to get it started ... I question if it will fly .... and will the
result still be a usenet like service? Why will people want to moderate the
traffic for such a big business for free?? Why should USENIX subsidize it?

Particularly if uucp/arpa/bitnet based usenet stays in tact as a cost free
competitor. I don't favor disbanding the technical communication within
usenet for what is likely to be an expensive, general public, mass marketed
data service. No matter HOW NEAT THE IDEA OF SATELITES SOUND.

To make stargate large enough to break make a profit, it will have to
target larger populations like the IBM, Apple, Atari, and other computer user
populations with a low cost BBS access competitve to Compuserve, etc.
It will take a long time to build the user base AND a lot of money.
With the usenet traffic opened up to such a large general population I question
the quality of the resulting service as a technical forum.

A lot more can be done to improve the cost of the usenet long haul connections,
which WILL LIKELY COST LESS THAN stargate.

	1) Upgrade the longhaul traffic to 9600/18000 baud or faster modems. The
	cost payback is several months, particularly if they can be purchased
	as a block buy with an agressive deal. I would guess that a group
	buy could get such modems near/below $1,000 each for a 100+ unit buy.
	I would be happy to coordinate such a buy.

	2) Upgrade the uucp server to be full duplex --- IE carry traffic in
	both directions concurrently -- this will likely improve the
	connection costs about 30% for backbones and have little affect
	on leaf sites.

	3) Implement a better I-have/I-want transmission scheme that is
	real-time. This alone could reduce phone traffic in the backbone
	by another 10% or more. With a full duplex communication channel this
	is really feasible.

	4) Negotiate a reduced flatrate DDS nite service AS A GROUP with
	one of the long haul carriers -- this could drop the costs another
	30% or more. This could be made a big PR deal with agressive bidding.

	5) evaluate X.25 major city interconnections with one or more of the
	major data carriers based on a flat group rate. This could be even
	cheaper than DDS. Again this could be made a big PR deal.


High technology like stargate is neat -- but I think we are just starting
another expensive data service by forging on past the experiment.

I think that USENIX should spend a matching sum to what it has on stargate
to evaluate alternate technologies and their implementation/service costs
before proceeding with stargate. I would be happy to act as a consultant
for such a project.

NOTE: please don't pick at the above loosely laid estimates .... but rather
present a COMPLETE and FEASIBLE estimated BUDGET for stargate service and
costs. ONLY THEN can we start to compare costs.

I am also starting a survey of current transmission costs. Please mail me a
detail of the following for your site(s):

	1) Your site name, sites you feed, sites that feed you.

	2) Transmission medium for each feed and what the cost of the
	   medium is ... if the medium is phone service, who is your
	   carrier, what are their rates, and what were your monthly
	   phone charges to each site for the last 6 months? What percentage
	   of the traffic do you/they pay for?

	3) What modem service do you currently use (1200/2400/9600?) and
	   what is the effect service rate ... check SYSLOG and estimate
	   by dividing bytes by seconds for news batches. Is your feed
	   currently compressed?

	4) Would you upgrade to 9600 DDS service to carry news?

	5) How many estimated news readers are at your site.


I'll post a summary to the net.
-- 

John Bass (DBA:DMS Design)
DMS Design (System Design, Performance and Arch Consultants)
{dual,fortune,polyslo,hpda}!dmsd!bass     (805) 541-1575