From: jim bell <jimbell@pacifier.com>
Date: Wed, 21 Aug 1996 14:53:41 +0800
To: vipul@pobox.com
Subject: Re: Phoneco vs X-Phone
Message-ID: <199608210046.RAA01785@mail.pacifier.com>
MIME-Version: 1.0
Content-Type: text/plain


At 04:02 AM 8/21/96 +0000, Vipul Ved Prakash wrote:

>Now lets see, you say we have enough capacity out there, alright, but then
>why is everyone raving about "a slow internet". We all know in IT 640K
>is never enough, niether is 640Mb. At some point in time, new cables
>_have_ to be laid. Moreover you seem to be considering a static growth rate,
>but we all know the Internet is nothing short of a big-bang. 
>
>I am trying to debate a model rather than numbers and in which case its
>important to to consider a long-term scenario.

Well, okay, here's some numbers: First, I've recently found out that in 
large quantities, it costs 10 cents per meter for bare fiber.

Assume 30 cents per meter per fiber for cabled fiber, or about $10 (US) per 
meter for 36-fiber cable.  Each fiber pair should be able to handle 
approximately 1 million conversations at current data rates, or a total of 
18 million conversations for that 18-pair cable, or 9.5 trillion 
conversation-minutes.

Multiply this cost by 10 for right of way, trenching, repeaters, and other 
auxiliary hardware, or $100 per meter.  This is probably just a ballpark 
estimate, but...

Let's assume that the average phone call goes through 5000 kilometers of 
fiber, which is approximately the width of the US.  (yes, I realize this is 
vastly over-inflated.  But my estimate is attempting to be conservative, on 
the "safe" side.)  A cable the width of the US costs  $500 million.  Assume 
that if it is to pay for itself, it must provide gross revenue of at least 
this amount to make a profit.

(It would probably be more accurate to say that the cable needs to actually 
profit by 10% of its value per year.  If it grossed 100% of its value, per 
year, then that would provide up to $450 million per year  in maintenance 
and other costs per year, and still be a profit of 10%.  But again, I'm 
trying to be CONSERVATIVE in how I account for costs.)

Assume an average of 1/10th of its capacity is used, or 950 billion 
conversation-minutes.  Do the division, dividing $500 million by 95 billion, 
and this works out to 1/20 of a penny per conversation-minute.

At this point, your jaw should drop.  Despite my dramatically over-inflated 
cost estimates for the fiber and installation and maintenance, and vastly 
underestimated utilization, I've still managed to justify only 1/20 penny 
per minute of cost. 


>This reminds me of a survey on a "proposed Rural Telecom Network" back in 
>India, which finally decided that ROI will not justify the project
>even in 20 years time.

I've read that estimates show that it would probably be cheaper to provide 
cellular-telephone service in China to everyone than to wire the country up 
with copper lines.  This isn't particularly surprising.  Cell-phones solve 
the "last few hundred yards/mile or two" problem quite well.  Since nearly 
all of the actual connections in a copperline telephone system are 
switch-to-individual-phone lines, going cellular saves a bundle of 
installation costs.
 

>> In the US, the current telephone company infrastructure is ALREADY PAID 
FOR. 
>>  It was paid for by over-inflated rates during a monopolized era.   If 
>> anything, the locals have an "unfair advantage" over the rest of the 
>> companies:  Only they have a copper pair into every home.  
>
>Alright, once again I try to show what exactly I am pointing at. Alice uses
>the phone 23 hrs a day, Bob uses 10 mins, both get their connections
>from ABCTel. The bandwidth with ABCTel saturates and it has to buy more 
>bandwidth (if ABCTel is a babybel, it will be buying it from a BigBell,
>and whether or not bandwidth exists is quite besides the point here)
>Its because of subscribers like Alice ABCTel is buying more bandwidth,
>and the flat rate revenues generated by addition of a couple of members won't 
>justify the new bandwidth. 


>Flat rates are based on the assumption that a subscriber will use the service
>for X amount of time. Since the phoneco has no control over user behaviour,
>more that reasonable number of Alice-type clients will screw up the phonecos 
>economics.

 Re-read my estimate, above.  Apparently, POTS-level "bandwidth" in a 
national fiber network should "cost" about 1/20 cent per minute, or maybe 
even less than that.  Your commentary just above shows that you are assuming 
a far greater cost.   For example, even your "23-hours-per-day" exaggeration 
should only cost 23x60x$0.0005, or $0.69 per day, or $21 per month.  High, 
but that usage is unrealistic, of course.  However, assuming one eighth of 
this usage, or 3 hours per day, that would be a daily cost of approximately 
$0.08, or about $2.75 per month.

Also, you need to remember that if the telephone usage (in for example, the 
US) was so high as to get a substantial fraction of the population on the 
phone, LD, for as long as 12 hours per day, the average utilization of the 
fiber cable I've projected would rise from the highly-conservative figure of 
10%, up to perhaps  25% or so.  Since this increased usage wouldn't increase 
the cost of the cable itself or the installation or maintenance, the 
corresponding cost estimate would drop from 0.0005 dollars per minute to 40% 
 of this or so, which is about $0.0002 per minute.  At that rate, a "12 
hours per day" usage would cost only $4.32 per month, which is well in line 
with current phone plans.

But again, remember that most of these "costs" are really merely the result 
of allocating some proportion of an existing, fixed cost.  Make that cost 
small enough (as fiber optics does) and spread it over enough users, and it 
becomes ignorable.


>With explosion of Internet, people have found a new way of using their phone
>line, and all these companies are already in trouble. To top all that we
>are loudly professing the Internet (ultra-low-cost) solutions to LD Calls
>with phoneco. 

I don't see any problem at all!  You're just (falsely) assuming that LD 
should, somehow, be more expensive than local calls.  Maybe it should be 
VERY SLIGHTLY more expensive,  but LD fiber is basically a one-time 
investment  with little maintenance costs.  

In order to determine how much more expensive LD should be over local, all 
you should have to do is figure out how much the additional equipment and 
fiber costs, and then consult an telephone engineer to figure out usage 
rates, and then go to an economist who can tell you how many more dollars 
you're going to have to charge.  (Based on typical ROI tables,etc.)   I've 
done enough of the numbers already to convince me that the amount is so tiny 
as to be ignorable.  


The fact is, LD phone is a business that, like it or not, is going to shrink 
drastically _in_terms_of_dollars_, simply because the cost of that service 
will likewise, go down.  That's life.


Jim Bell
jimbell@pacifier.com