g-leech / argmin-gravitas Goto Github PK

Mr Argmin Gravitas?

What's interesting to you 👌

Hello,
I notice a few missing or erroneous entries for your "data" (the spreadsheet with relevant information) I thought I should clear them up most are for the IBM Selective Sequence Electronic Calculator.

In terms of "Protagonists" (heading from the google spreadsheet) for the SSEC you list A Wayne Brooke, when you bring it up under the webpage/app, the IBM SSEC is listed as by A Wayne Brooke & Wallace Eckert. A Wayne Brooke was the engineer heading the team running and maintaining the SSEC during its operation, not a head designer (I think he may have had a subordinate role on the design team). I think the official IBM announcement credited Wallace Eckert as one of three chief designers, however accounts of the design process suggest that while he was the internal customer who signed off on the design the actual design work was done by Francis E. "Frank" Hamilton and Robert "Rex" Seeber. So probably Hamilton & Seeber should be for who the SSEC is by and Wallace Eckert also had a key managerial role in the machine's construction. Seeber is credited with being the one who advocated that the SSEC instructions be in the same format as data and so allowing the SSEC to use self-modifying code, one definition of the stored program concept. If the "by" is supposed to reflect a more fluid concept then Eckert is certainly appropriate as a main representative of the machine and Brooke is significant both as someone who gave interviews about maintaining the machine (he called himself the SSEC's psychoanalyst) and drafted but never published a defense of the machine. For documentation of this see IBM's Early Computers pages 48-51, also this website collects some useful material and references: http://www.columbia.edu/cu/computinghistory/ssec.html

The SSEC did fixed point arithmetic, the spreadsheet has that line blank (although I don't think that column of data is used by the web app?).

For Base you list the SSEC as being base 2, the SSEC was a binary-coded decimal machine, therefore some aspects of its arithmetic and storage would be decimal, but other aspects of the way components acted would be binary. I suppose this data is not wrong but rather the category is maybe underspecified depending on the purpose. There are a lot of potential things to track, I suppose this is up to the designers discretion.

For Bit transmission you have it as serial, in fact the IBM SSEC transmitted numbers all 78 bits (19 binary-coded decimal numbers and sign) at once from the Appendix of IBM's Early Computers (page 586) we have this description "The overlapped transfer of so many twenty-digit numbers or instructions required an elaborate system of buses. Comprising eight storage-out and eight storage-in buses, each having eighty conductors (seventy-eight for data and two for control)" (note often the number/word size of the SSEC is described as 20 digits as here but it is actually 19 digits plus sign). The SSEC had to transmit all its numbers quickly because the relays and paper tapes it used for memory and storage were so much slower than the electronic arithmetic unit that they minimized delay on them. Whoever said the Whirlwind was the first parallel bit transmission computer probably did not include the SSEC on the list as a computer.

A noticed a few things for the ENIAC:
For "Protagonists" the data on the Spreadsheet is "Wallace Eckert" but when you bring up the ENIAC on the web app you get John Mauchly and Presper Eckert. So the spreadsheet is not specifying the data that is actually used there.

For "Bit Transmission" for the ENIAC the spreadsheet cell is blank. I'm pretty sure the ENIAC used parallel channels to transmit data. Just for instance in the book The ENIAC in Action (page 237) we have the following passage about designing the EDVAC as a successor that suggests the ENIAC was a parallel machine "Its aggressive target cost placed a premium on simplicity, which pushed Eckert toward minimizing the logic circuitry by transmitting digits serially rather than in parallel." Again whoever said the Whirlwind was first probably meant of a list of machines that did not include early odd cases like the ENIAC.

A note for the Harvard Mark I (IBM Automatic Sequence Control Calculator)
For "Program Control" the cell in the spreadsheet is blank. In fact the Mark I was controlled by instructions on punched paper tape. This may not matter for the web app since the relevant criterion is just "stored program" or not "stored program", but given your entries for other machines do specify paper tape I thought it worth pointing out.

To prevent spam comments. Related issue: eduardoboucas/staticman#175.