VISIT TO COMPUTER CONTROLS, AVCO AND ITEK

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP80B01139A000200010004-7
• Release Decision: RIPPUB
• Original Classification: U
• Document Page Count: 4
• Document Creation Date: December 15, 2016
• Document Release Date: March 18, 2004
• Sequence Number: 4
• Publication Date: July 8, 1958
• Content Type: MF

Approved For Relea2~:
- Q 1139A000200010004-7
C?DIW-D-11
8 J1t],y 1958
INTELLIGENCE ADVISORY COMMITTEE
' C Into IAC Cif on Docummtatica
SUBJ HT
o Visit to Computer Controls, AVO and. ITEK
3 a hid mewarenth n is circulated for r intonation
by dire?ctton of the o
At t
25X1
Approved For Release 2004/03/31 : CIA-RDP80B01139A000200010004-7
 ApproFor Re~yt43~$:~CDP81ft02(10004-7
C
f
p
y
2 July 1958
ME 10i7ANDtfl4 FOR: Assistant Director, Central Reference
SUBJECT : Visit to Computer Controls, AVCO and ITEK
on l July 1958, in company with
I visited subject companies in Boston.-
1, is detrained at Route 128 Station and were met by Rr, Beni
Kassel of Computer Controls. After breakfast, we went to CC which
is located in Babson, Park, ~;ellesley, Maas9 kea strolled through
the plant looking at the manufacturing process for the transistorized
units to be used in electronic gear and especially the icfl iCARO
selector for OCR. The miniaturisation of this unit has reduced its
size from 6 x 8 feet to 4 x 6 feet, its unit power requirement from
120 watts to 3-4 watts, its air conditioning requirement to nil, and
its maintenance problem virtually to minimum. At the same time its
selection logic has been materially increased. A briefing by members
of the staff on the Air Force Document Data index device followed.
The Document Data Index Set AN/GSQ=26 is a f1e ib1e, high speed
information storage retrieval system. The information storage medium
is magnetic tape. Each reel of magnetic tape can contain up to 35
million bits of stored information:. The stored information. can be
searched at the rate of over 180,000 bits per second to seek the
answers to as many as ten questions simultaneously. Each of the
ten questions can be composed of any combination of fifteen phrases.
Each phrase can be made up of any combination of twenty question
words. Its designed chara::toriatics are:
a The memory unit is 1 inch magnetic tape with 2 alpha-
numeric characters, 2 parity bits and 1 timing bit
across the width of the tape. The packing density
is 220 pulses per inch. The tape is scanned in either
direction at 1.10 inches per second or an entire reel
in 4j minutes.
b~ An average document of 20 words, of 7 characters per
word, appears to consist of document number and such
other descriptives as the Air Force desires.
The tape is set up in blocks averaging 7.3 documents
per block or a block length of 2.66 inches plus 0.,4
inches of gap. This gap permits the tapC to be
stopped between blocks when a hit is made. About
25X1 25X1
Approved For Release 2004/03/31 : CIA-RDP80B01139A000200010004-7
 Approve or Release T4.103131j : CCIA-RDP80 0 139A000200010004-7
Its # r USE ONLY
2(0 documents per second are seamed, and there are
about 638500 document8 per real.
d. The first operational mode is the input from punched
paper t ,~"pe to magnetic tape. The paper tape is read
into a core buffer until 1092 characters have been so
stored then the buffer dumps onto the magnetic tape.
The input rate is 200 characters per second or 1,3
documents per second or 300 documents in .3 minutes.
o. question words are inserted into the high speed internal
memory of the machine by means of punched paper tape.;
Logical assembly of the words into phrases and of the
phrases into questions is specified by means of plug-
board connections. The plugboard provides for
specifications of equality, "greater.than% or "less-
than" comparisons between stored and question words,,
and also makes both assertions and negations of words
and phrase comparisons available for incorporation
into logical statements of the phrases and questions,,
f. Selected documents may have 3 words printed out,, one
of which must be the document number. Again the hits
are registered in the core memory and printed out on
a teletype receiver or a paper tape is made and fed to
* F'lexowriter.
g. An editing capability is being designed into the machine
so that documents may be erased, however this process
looks unproductive and uneconomical.
In short,, the machine is a mechanized index to a large document file?
I believe we should keep close watch on the development of this device
for possible utility with proven MINICAM gear (camera, processor, etc.)
as an index to a 111, UCARL filet,
2~ be went next to A'CC' where Messrs. Martin and Phaneut described
and showed us a model of their idea for document storage, In essence
the idea is:
as Photographs of documents are stored at 100x min:.if.i.cation
on glass panels, 10,.00,0 pages per panel..
b. An optical-.electronic readout displa*s the document on a
TV tube. Under present operating practice, the document
Approved For Release 2004/03/31 : CIA-RDP80B01139A000200010004-7
 Approved For Release 2004/03/31 : CIA-RDP80B01139A000200010004-7
FOR C.FFICIAL US1 ONLY
-A3m
could be read only with great difficulty owing to a one
second sweep down the tube. Resolution at all points on
the tube is good*
c. Spgcific documents would be located by plate number and
x and y coordinates.
1:e were all impressed with the :tube Goldbergish aspect of the device,
and the probable maintenance problems with so many delicately adjusted
electronic components which would have to move quickly and stop
suddend],y. In short, we thought the idea interesting but in its
present operational attitude impractical..
3, Our third visit was to ITEK where Dr. Art Tyler gave us a
tour of the optical laboratories in townjthen we went out to the
plant. The most interesting development in town was an aerial
camera capable of sweeping an are of 1200 in minimum time, and an
8x enlarger which created a strip print of the sweep. At the plant
we saw a high fidelity viewer and scanner for aerial photographic
transparencies. Later we met with Dick Leghorn to discuss ITEKfa
future plans. They propose to join the throng in the information
processing field and possibly in mechanical translation. One subject
in,the former field of interest was the mechanical definition of
questions, i.e., a means to assist the information service to define
precisely what its customers want. I rientioned character reading
as being of great inter t in several applications. Dick Leghorn
commented that the pendum seems to be swinging away from image
and coding storage on the same medium, but rather toward separate
index for subject or area searching and storage of the images. The
advantages seem to lie in file management.
4. Our formal meetings broke up a few minutes after 1800 hours,,
but Art Tyler joined us for dinner at the Red Coach Inn for more
philosophizing on the subjects of the day.
Deputy Assistant Director
Central Reference
Approved For Release 2004/03/31 : CIA-RDP80B01139A000200010004-7