COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	.require "let.pub" source file
C00011 ENDMK
C⊗;
.require "let.pub" source file
∂MLOT William Miller$$Editing Use of LOTS and other matters∞

Dear Bill:

	Ed Feigenbaum told me that you expressed uneasiness about the
use of LOTS for pure text editing.  Here is the situation as I see
it.

	1. As I told you a year ago, I am an enthusiast for text
editing and hoped to encourage it as soon as it became clear that
LOTS was successfully meeting its other obligations.

	2. This condition has not been met and cannot possibly be
met any sooner than the end of Fall quarter, since that quarter will
be the first in which LOTS undertakes to satisfy a full course load.

	3. I was mistaken in thinking that editing of papers and
dissertations would require encouragement.  The problem is whether
and how to suppress it.  The use started spontaneously, a number of
PhD theses and other documents were prepared on LOTS now.  In checking
the new users who signed up in the last month I noticed one
student from the English department put "thesis" as his purpose in
using LOTS.  It is hard to imagine that his intent can be much other
than editing and this shows how far this idea has spread.

	4. I doubt that editing was an important cause of overloading
in the Spring quarter even though at some times I noticed that about
a quarter of the terminals were editing something other than programs.
I believe that the editor does not put heavy demands on the system.
However, this information is not certain, because the operating system
doesn't provide information about how much one user is competing with
others.

	5. Towards the end of Spring quarter I was tempted to try
to suppress editing but refrained, because there had been no warning.
For Fall quarter, there will be a notice that load-shedding measures
may be required if there is a crunch and that text-editing will be
the first to go.  In particular, anyone who expects to complete a
term paper on LOTS runs the risk that he will get cut off just before the
end of the quarter if the load of running programs for courses is too heavy.

	6. When we have met our other obligations, I will be back to
you with plans and costs for encouraging text editing.

	7. I recognize that we promised you a document on the costs
of text editing.  I haven't had time to do it myself, and the staff
has been overloaded with trying to make the system work better to
handle the load.  We'll get to it.
Any long range plans for editing will have to be creative about how to
pay for it.

	8. As you know, our equipment is scheduled for delivery at the
end of September.  I think we'll be able to put it into service promptly
without too much thrashing so that there is no need to ask CS105 and CS106
to use the 168 again.  However, there is not much room for bad luck
or for adapting to surprises about the behavior of the new operating
system we expect to receive from D.E.C. about the same time.

	9. The one bit of information about the costs of text-editing
I can offer concerns the cost of disk storage.  We got a package deal
that included the two new disk files - I think that apart from the
package, I would have ordered only one more.  Anyway we can't say exactly
what they cost us.  The list price is about $35K per disk and my guess
is that we got them for $25K which is still twice what the AI Lab pays.
Dividing the price by 50 to get a monthly cost and considering that a
disk contains 200 megabytes, we get $2.50 per megabyte month.  A very
large thesis would be about a megabyte, so this is an estimate of
the marginal storage cost.  I don't yet have good estimates of the
terminal costs or the CPU costs.  The terminal costs are computable
estimating $1K per terminal amortized over 25 months giving $40 per
month and estimating 400 useful hours per month gives $.10 per hour
terminal cost.  A student might spend 300 hours typing and editing
his thesis giving $30.00 for that.  As I said the CPU costs are harder
to allocate.  The share of the personnel and other costs that should
be allocated to editing is even harder to estimate.  It should be
considered low, because we have put no staff time into improving
editing or printing.  We also haven't had to increase our printing
facilities.

	There are two other ways of estimating editing costs.  If it
were true, as Feigenbaum indicated that you thought it might be, that
the expansion had been forced to meet editing demands, then the cost
would be rather large, but I'm sure this isn't so.  Another way is to
prorate the average cost of LOTS per terminal hour, about $.75 including
everything.  Then my hypothetical 300 hour thesis would cost $225,
but I don't believe that either.

10. When we get to encouraging editing, an important possibility
is a smaller computer dedicated to editing.

11. During the summer, the research and thesis use of LOTS increased
steadily.  The top ten users got 1/3 of the computer time, but I'm
sure they will claim that they used it late at night when no-one
else was competing for it.  The next 40 users got about another 1/4
of the time.  If nothing were done, this use would grow linearly with
time.  Therefore, the first task we shall undertake concerning controlling
usage is to see what the big use was during the summer, and check its
compliance with the LOTS rules.  Then the Advisory Board will recommend
changes in detailed policy to optimize use in accordance with
general policies.

	I hope this information will reassure you that we haven't
unilaterally encouraged editing and that we are more or less on
top of the problem.

CC:  Ralph Gorin, Maurice Bizzarri, Jim Adams