Massachusetts Institute of Technology, Tapes of Tech Square (ToTS) collection
Scope and Contents
The analog material mentioned above consists of four boxes of content including paper server/tape dump logs and contextual material including handbooks for using ARPANET, other servers, and timesharing systems.
The digital tape images and extracted files contain numerous directories with research and working files primarily textual in nature—code, software, logs, documents, emails, etc., organized by individual users or groups within these labs or by topics. Additionally, there are files created by non-MIT “guests” who had access to the MIT timeshare machines through MIT community members or networked environments such as ARPANET.
Included is an early versions of the source code to the interactive fiction game Zork from 1977 and January 1978, extracted for researcher ease of access.
- Majority of material found within 1970 - 1995
- 1960 - 2017
- Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory (Organization)
Conditions Governing Access
Intellectual Property Rights
The MIT Laboratory for Computer Science (LCS) was a laboratory for research in computer science and engineering, first established on July 1, 1963, as Project MAC in the Department of Electrical Engineering. By 1967, MAC had separated from the Electrical Engineering Department and started reporting directly to the MIT provost as an interdepartmental lab. The name Laboratory for Computer Science was chosen in 1975 and officially recognized in 1976. In 1981, while retaining its interdepartmental status, LCS came under the aegis of the School of Engineering, and the LCS director began reporting to the dean of engineering. The MIT Artificial Intelligence Laboratory (AI Lab) was established in 1959 and was a separate lab from LCS, focusing primarily on artificial intelligence. On July 1, 2003, the Laboratory for Computer Science (LCS) merged with the Artificial Intelligence Laboratory (AI) to form the CSAIL.
The files in this collection come from backup tapes of computers and systems used by these labs. Below are short descriptions of these systems, their acronyms, and context around how and when they were used:
ITS (the Incompatible Timesharing System) was the operating system written at the AI Lab and LCS to run on Digital Equipment Corporation computers of the 36-bit family; these were some of the DEC PDP (Programmed Data Processor) models. ITS was named by its authors in contrast with CTSS (the Compatible Time-Sharing System) running on the IBM 7094 system then at MIT.
The AI Lab's first such computer was a PDP-6. This was succeeded around 1970 by a PDP-10, KA model, the AI-KA. LCS also acquired a couple of KAs in the next few years, DM (mostly used by the Dynamic Modelling Group) and later ML (primarily used by the MathLab Group). It should be noted that the DynaMod Group used a somewhat different dumping scheme for DM than was used for all the other ITS machines. DM was also the only ITS machine (until the middle 1980s) to have a 9-track tape drive; all the others had 7-track drives.
When the Macsyma program was ready for somewhat general use by its sponsors (mostly US government agencies), the MC (Macsyma Consortium) machine was acquired for their use of it. MC was a later, KL model of the PDP-10. During the last year of its life at MIT, the MC-KL was renamed MX for technical reasons. DEC had written its own operating system for the PDP-10, TOPS-10 -- but so had other CS research centers. In particular, the hackers at Bolt, Beranek, & Newman wrote what they called the TENEX system. DEC later acquired rights to TENEX, and used it as the base for its next PDP-10-type operating system, which it called TOPS-20. This was written for DEC's next model in its 36-bit family, the DEC-20, which under its skin was remarkably similar to a DEC-10-KL (same CPU, different memory and I/O channels). Hackers who knew its history regularly referred to Tops-20 as TWENEX.
LCS acquired the XX DEC-20 machine in the late 1970s, and the AI Lab acquired OZ in the early 1980s -- in each case to begin to shift its research groups off the ITS machines.
The KAs limped along into the early 1980s before being shut off, and MC continued through 1986 -- when some new, physically smaller DEC-20s arrived. These were of the KS model, and ITS was adapted to run on them (this was mostly a matter of rewriting the microcode). The AI, MC, ML, and MD KSs were in operation for the later part of the 1980s, although only the AI and MC machines were much used. In fact, it was intended that the MD machine could be shut down and a version of TWENEX brought up on it – in which case it would be referred to as MIT-LSD; this happened no more than a few times. Separately the labs also acquired a 5th KS, which did run TWENEX as its regular operating system; this machine was named BLT (Brave Little Toaster) after a science fiction story.
The AI Lab and LCS DEC-20s lasted until the end of the 1980s, when the labs fully shifted to smaller, per-group and later personal computers running varieties and descendants of Unix.
 “Intro-to-ToTS,” Massachusetts Institute of Technology, Tapes of Tech Square (TOTS) collection, MC-0741. Massachusetts Institute of Technology, Department of Distinctive Collections, Cambridge, Massachusetts  “MIT Leaves behind a Rich History in Tech Square,” March 17, 2004. https://news.mit.edu/2004/techsquare-0317.  “Biographical/Historical Note.” Massachusetts Institute of Technology, Laboratory for Computer Science records, AC-0268. Massachusetts Institute of Technology, Department of Distinctive Collections, Cambridge, Massachusetts. Accessed April 23, 2020. https://archivesspace.mit.edu/repositories/2/resources/228.  “Mission & History.” MIT CSAIL. Accessed April 23, 2020. https://www.csail.mit.edu/about/mission-history.
1.3 Terabytes (13,424,283 digital files)
3.33 Cubic Feet (3 record cartons; 1 letter manuscript box)
Language of Materials
The digital files remain as organized in the ToTS project area on the MIT Athena system. The files and access tools therefore expect a UNIX-style file system and layout.
The raw tape images are in the subdirectories “/tots/images” and “/tots/part1” through “/tots/part8”
The original versions of the extracted directories and files are under the “/tots/www/data/extract/” directory.
The contents of the /files/byname subdirectory provides access to the files, sorted by the name of the directory that contained them, via links into “/tots/www/data/extract/”
De-duplicated extracted trees live in the “/tots/its20x” subdirecotry. CSAIL indicated that this is the best version of the extracted data to use.
Links and subdirectories aek-progs, snapper, tapetools, tools, and tulz contain or refer to various software tools CSAIL used to produce the extracted files.
More information about different directories and sections of the data can be found in the Intro-to-ToTS, README, and ROADMAP files in the top level of the “/tots” directory.
Immediate Source of Acquisition
Researcher Lars Brinkhoff extracted the Zork 1977 files from the 9005196.tap tape image file and the Zork January 1978 files from the 9005201.tap tape image file within the /tots/recovered/vol2 directory using the itstar program
Below is a brief outline of steps taken by the ToTS project team in CSAIL before transfer to DDC.
In the early 1990s, LCS/AI Lab members read a selection of the early 7-track tapes written on the labs' ITS machines and rewrote them onto new 9-track tapes at a higher density. In 2003, they began to have all the tapes read and transferred to accessible storage. Special methods had to be used to read the data on the tapes without scraping the flaking oxide (which contained that data) off them. CSAIL did not have the technology to do this reading, so they outsourced this work, mostly to a specialized facility in Canada, producing image files which then had to be further processed.
Each tape transfer produced an image of the bits and a photo of written or printed tape labels on the physical tape reel. Image files have been given names like “1234567.tap,” photos names like “1234567.jpg.” According to project staff, old tapes have yielded many read errors, some photos are missing, and there are more errors, omissions, duplications, etc. still undetected.
CSAIL ran the image files through specialized programs to decode them back into the directories and files that had been written onto the tapes, as well as some of the metadata concerning each tape, focusing on the ITS and Twenex tapes.
Due to the nature of backup tapes, numerous files were duplicated on many of the tapes involved. To save space, CSAIL extracted directory trees and compressed them, replacing multiple copies of each identical file with links to one copy of the file. These de-duplicated extracted trees live in the “tots/its20x” subdirectory. CSAIL indicated that this is the best version of the extracted data to use.
More information can be found in the Intro-to-ToTS, README, and ROADMAP files in the top level of the /tots directory.
- Artificial intelligence Subject Source: Library of Congress Subject Headings
- Massachusetts Institute of Technology. Artificial Intelligence Laboratory
- Massachusetts Institute of Technology. Laboratory for Computer Science
- Massachusetts Institute of Technology, Tapes of Tech Square (ToTS) collection
- Joe Carrano, Greta Suiter, Kari Smith
- Description rules
- Describing Archives: A Content Standard
- Language of description
- Script of description
Part of the Massachusetts Institute of Technology. Libraries. Department of Distinctive Collections Repository
Massachusetts Institute of Technology Libraries
77 Massachusetts Avenue
Cambridge MA 02139-4307 US