COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	sail[f81,jmc]		The finances of SAIL
C00009 ENDMK
C⊗;
sail[f81,jmc]		The finances of SAIL

To: whom it concerns

Facts:

1. The marginal cost of running SAIL is about $200K per year.

2. This can be reduced by reducing personnel.

3. SAIL's share of the present CSCF budget is close to $400K per year.

4. The present SAIL users don't have $400K per year but they do have
more than $200K to spend for computing.

5. SAIL is not very heavily loaded and could take more users if the
increased crowding were seen to be worthwhile in order to keep costs
down.  Of course, SAIL is slow in the middle of the afternoon, but
I have found that one can not only edit effectively but can run many
large programs like EKL and can also debug small LISP programs.
At other hours SAIL is quite unloaded.  Of course, people doing AI
will always push their computer facilities to the limit and want
more.  Nevertheless, given the present balance between cost and
performance, SAIL isn't heavily loaded.

6. Because of its non-standard software, SAIL is seen as less desirable
than SCORE by new users.  Many old users prefer the features of SAIL.

7. If SAIL could be replaced by a 2060 with similar
memory, disk storage and terminal capacity today without cost, this
might be the right thing to do in spite of the loss of features.
However, this replacement could not be done for less than a million
dollars.

8. SAIL is a software dead end in spite of the fact that its software
has many features more advanced than exists elsewhere.  Because these
features were not written up, the world has moved in other directions.
The department needs to determine how these features, line-editor and
more recently operating out of E, efficient display and real-time service,
and enlarged character set can be obtained in its new machines.

9. Closing down SAIL is not a reasonable option until the 2080 or
S-1 becomes available to give several times the performance at a
cost comparable to the present cost of replacing SAIL by a 2060.
This will be at least two or three years depending on when these
things become available and the time it takes the Computer Science
Department to get the money.

10. The present formula for allocating charges won't work, because
SAIL users don't have the money to pay their allocated share.  It
is also somewhat unfair, since most of SAIL's allocated share of
the general expenses are going to pay for new developments only
a small part of which directly benefit SAIL users.

11. CSCF should allocate charges so that SAIL pays for itself.  Lower
charges on SAIL are probably needed in order that new users feel
compensated for learning its non-standard features.

12. There are some genuine choices that can be made concerning what
efforts should be made to get more business for SAIL.  In the first
place, maybe this is unncessary.  Since present usage will pay for
more than operating costs, a rate can be found that makes SAIL pay
for itself.  If this rate is less than that charged by SCORE, some
usage will migrate.  (I am assuming the correctness of Ralph's
assertion that SCORE can sell all the aliquots it has).  On the other
hand, I have had no difficulty in attracting mathematicians
and philosophers with whom I have had personal contact, and they
continue to find SAIL cost-effective for their purposes without any
further assistance from me and perhaps from anyone else.
Probably the smallest sales campaign would attract people who have
been convinced by the CONTEXT sales campaign but can't afford the
CONTEXT rates.

There is one software modification that would make this more attractive.
Namely, if a version of E were made to work with Zenith, etc. terminals
or if EMAX were made available, then users who can't afford terminals
with the SAIL character set could work effectively.  Perhaps they
already can, and the fact needs only to be properly documented.

The considerations determining this choice should be

(1) Do we need more users to be sure that SAIL can pay its operating costs?

(2) Assuming we don't actually need more users, do we want more users
anyway, so we can help pay for the Ethernet development costs and
perhaps some of the File Computer development costs?

This is a matter for the Department to decide after consultation
with all interested parties including SAIL users and the CSCF
managers.