At MIT's Artificial Intelligence (AI) Lab and Laboratory for Computer
Science (LCS), we have rooms full of 7-track and 9-track magnetic
tapes in various states of decay. These tapes are the incremental,
full, and archival dumps of all the machines used for everyday work by
students, faculty, and staff in the Labs from
1971 to 1990. The data on any individual tape may or may not be
valuable; we are unable to tell. Much of this data was written by
operating systems no longer in common use, such as ITS, TOPS-20, and
Genera. An air conditioner malfunction has allowed water to leak onto
some of the tapes. Some of the protective outer tape rings have grown
old and brittle. Many have already failed, causing the tape to fall
off the tape racks and their cases to shatter. Many tapes are being
stored off-site in an environmentally controlled warehouse, but some
are completely un-cataloged. Others may be sitting lost in the lab's
basement somewhere. Furthermore, we are losing the knowledge required
to make sense of the data on the tapes, which could be in one of a
dozen formats. Even perfectly preserved tapes may thus soon be
reduced to gibberish. We need to pay attention to more than just the
tapes themselves; we need to rescue data from the operators who wrote
these tapes and from the paper dump logs. In summary, we are losing
irreplaceable data every day.
Operating System | Machine | Density (bpi) | Format | Approximate Year Written | Number | Size | |
---|---|---|---|---|---|---|---|
ITS | DM KA-10 | 800 | 9 track | 1976--83 | 330 | -- | |
ITS | AI KA-10 | 800 | 7 track | 1971--82 | 362 | -- | |
ITS | ML KA-10 | 800 | 7 track | 1973--83 | 203 | -- | |
ITS | MC KL-10 | 800 | 7 track | 1976--85 | 511 | -- | |
ITS | MC KL-10 | 1600 | 9 track | 1981--86 | 60 | -- | |
ITS | All KS | 6250 | 9 track | 1985--90 | 159 | -- | |
ITS Total | 1625 | 77GB | |||||
Tops-20 | AI OZ | 1600 | 9 track | 1982--86 | 397 | -- | |
Tops-20 | AI OZ | 6250 | 9 track | 1986--88 | 118 | -- | |
Tops-20 | LCS XX | mixed | 9 track | 1976--88 | 423 (in storage) | -- | |
Tops-20 Total | 938 | 132GB | |||||
Mixed/Unspecified | 1600 | 9 track | -- | 757 | 30GB | ||
Mixed/Unspecified | 6250 | 9 track | -- | 436 | 63GB | ||
Grand Total | 3,756 | 302GB |
The long-term goal of this project is to provide a usable system
whereby users can easily search and access all the Labs' archival data
in a uniform fashion, and to preserve this data in such a manner that
it can be decoded even if all the supporting documentation for it is
lost. The more immediate goal is to assemble the incremental, full,
and archival backups from the Incompatible Timeshare System (ITS),
record them on new media in a long-term (archival) format suitable for
use well into the future, and provide tools to help search and manage
this large data set in an effective manner. The entire collection of
readable ITS backup tapes contains approximately 77 gigabytes of data,
and constitutes our smallest collection of tapes.
Table provides a brief breakdown of the tapes that
have been accounted for and are believed to contain useful data.
There are other tapes around the lab which are unlabeled, and there
are many tapes known to be uninteresting ``scratch''
tapes.
Starting in 1968, ITS was the workhorse of Project MAC, which later
divided into the Laboratory for Computer Science (LCS) and the
Artificial Intelligence (AI) Lab. ITS ran on a two hardware
platforms, the DEC PDP-6 and later the PDP-10; and was implemented
entirely in MIDAS, an assembly language with macro
facilities
There has been a demonstrated need for some of the data on these tapes. We are not just trying to rescue data that will never be used. Here are some examples of requests that we have encountered: