Numeric vs Alphanumeric Login: Which came first?
Please share any information you have about numeric and alphanumeric logins/ID's.
Numeric IDs preceding Alphanumeric Les Earnest
Why early logins used numbers instead of names Les Earnest
Choosing 'badge numbers' as IDs for LTSS Norman Hardy
Other timesharing systems that supported logins Jim Babcock
Les Earnest
Text taken from this Original Post 19 Feb 2002
Amy also asks:> What was the first alphanumeric user ID?
> – Did numeric precede alphanumeric?
Based on the preceding discussion, numeric user IDs preceded alphanumeric
for general purpose timesharing systems.
The first first commercial timesharing system was the DEC PDP-6 in
1965. Like CTSS it used numeric "project" and "programmer" numbers, but
stored as binary numbers between 1 and 2^18, and passwords were alphanumeric.
Note that users had functioned as assistants to computers for many years,
supposedly because the latter were so expensive. When the introduction of
timesharing put users on a more equal footing with machines they were
identified by number rather than name. I leave it to psychologists to
figure out whether this was done to save a few precious bits of memory
(though not in CTSS) or whether it was to keep users in their place.
Jerry Saltzer also says:
>We didn't begin using strings of alphabetics as login identifiers until
>Multics came into service. But by then the concept of logging in with a
>user id and a password was widely appreciated, so one might have to do a
>survey of the time-sharing systems extant between 1963 and 1968 to figure
>out who deployed alphabetic login identifiers first.
I don't know when Multics became operational but upon delivery of a PDP-6
system at the Stanford Artificial Intelligence Lab (SAIL) in July 1966 we
promptly hacked it to accept alphanumeric identifiers of up to three SIXBIT
characters, which fit in the same space as the old numeric ID. Many of us
still use three letter IDs today.
Another candidate for the first alphanumeric login was ITS at the MIT AI
lab. I don't recall when it became operational but suspect that it was
later than SAIL.
However, before the first timesharing systems became operational there was
at least one file system that used alphanumeric user IDs. This was a
reel-to-reel tape drive and accompanying software that was used to store
files on the TX-2 computer at MIT Lincoln Lab. Tom Stockebrand, who
designed and built it, recalls that it began operating around the Spring of
1961. Both user IDs and file names were alphanumeric, stored as SIXBIT (I
was LES), and were probably up to six characters long, given that it was a
37 bit machine. A timesharing system was later developed for TX-2 (circa
1966) and, I believe, continued using alphanumeric user IDs.
Thus it appears that the first use of alphanumeric user IDs was on the TX-2
file system in early 1961 and the first use in a general purpose
timesharing system was either in Multics, SAIL (whose TS system was later
named WAITS), ITS or TX-2 TS, whichever came first. I trust that we will
be able to sort this out before long.
Les Earnest
Text taken from this Original Post 24 Feb 2002
Responding to my conjectures about why early logins used numbers instead of
names for user IDs my former colleague, the pseudonymous greep, wrote:
[Original Post 21 Feb 2002 greep wrote that the CPU time required to look up an alphanumeric identifier took longer that of a numeric identifier. A numeric identifier could be used as an index into a table. CPUs were much slower then and efficiency was important.]
A valid point. In order to do an efficient lookup on a name it is
necessary to use some form of hash coding. I don't remember that being
used much, if at all, in the early 1960s.
greep continues:
[greep feels that people who worked with IBM systems considered alphanumeric IDs as "too easy" when computers were supposed to be hard to use.]
Also true. Furthermore the idea of assigning ID numbers to people has
always been popular with governments, large employers and fascists.
Throughout the 1950s computers were used almost exclusively for accounting
and numerical analysis, hence there was a predisposition to regard all
information as numeric. I observed an example of this during my first job
as a programmer in 1954-56 at the U.S. Navy's Aeronautical Computer Lab in
Johnsville, Pennsylvania.
Our lab's responsibility was to do flight simulations of missiles and
manned aircraft using the world's fastest computer, an electronic analog
machine called Typhoon that had been built by RCA. However, because there
were many ways that Typhoon could go astray we also ran digital check
solutions using an electromechanical computer called the Card Programmed
Calculator (CPC), which had been created by Northrop Aviation out of a
bunch of IBM accounting machines. Everything was programmed in decimal
absolute and the accounting machine that printed results could print only
numbers and a little punctuation.
Near the end of my Navy tour a decision was made to switch to the IBM 650,
which was a much faster and more reliable magnetic drum machine. The
cheaper version came with a numbers-only printer but for a slightly higher
rental there was an alphanumeric printer, which was required if we wanted
to use the assembly program called SOAP. SOAP was available for free, like
nearly all software in that era, and not only enabled programmers to assign
mnemonic names to variables but also automatically optimized memory
allocation taking into account drum latency. However, the management
decided to continue using a numbers-only printer, requiring programming to
be done in decimal coding. Never mind that this greatly reduced programmer
efficiency.
In summary, there were at least four plausible reasons why early
timesharing systems used numerical user IDs rather than names:
1. It saved a few precious bits of memory (though apparently not in CTSS),
2. It speeded up login processing time slightly,
3. It was consistent with the earlier numbers-oriented computer culture,
4. It kept programmers in their place by not acknowledging that they had names.
Regarding the question about when alphanumeric IDs first appeared in
logins, that is still a horse race. In my preceding note I listed a number
of candidate systems for that honor but I later realized that I had omitted
quite a few. In my next note I plan to catalog them and invite readers to
help me fill in some of the blanks so that we can pick the winner.
Norman Hardy
Original Post 22 Feb 2002
"I remember choosing "badge numbers" to serve as user IDs for LTSS,
(Livermore TImesharing System) about 1963..."
Jim Babcock
Original Post 25 Feb 2002
Babcock lists several other timesharing systems that supported logins.
Link to this Page