Mel's Loop - A Comprehensive Guide to The Story of Mel

When I read the Story of Mel, I had not heard of the LGP-30 or RPC-4000 elsewhere, so thought the computers were not particularly notable. (I think it’s that they were small computers, so user groups and culture didn’t form around them in the way they did around large mainframe computers.)

But the LGP-30 was used by Edward Lorenz to discover chaos theory and the butterfly effect and all that.

Lorenz’s software was initially written by Margaret Hamilton, but the custom at the time was not to give co-author credit to female research assistants, so she is often left out of the story.

I’ve talked about this before, but many years ago, I reached out to Professor Nather (who wrote the Story of Mel) and he and I corresponded a bit. It’s all true, all of it.

If I can find the email, I’ll paste it here, but it’s easily 25 years old at this point…

Previously, The Story of Mel, annotated (1996)

ah! I’ve been looking for this story for years, I could never remember the title or the guy’s name, thanks for sharing this great annotated version!

This is exactly what I hoped it was from the headline! I remember the story of Mel from the early 90s on Usenet, by which time it was already old folklore. Thank you for sharing this fun read here.

The key part in this post is the “missing bits” part: https://melsloop.com/docs/the-story-of-mel/pages/mels-hack-the-missing-bits

I loved the story since I was very young, but there always seemed to be a contradiction in saying “every single instruction was followed by a GO TO” and “The carry would add one to the operation code, changing it to the next one in the instruction set: a jump instruction.” Why would a system where every instruction encodes the next instruction bother implementing a jump instruction?

Reading the “missing bits” part is enlightening and I like the way it’s written to avoid throwing shade at anyone. Note that the Story of Mel was written over 20 years after the events occurred. Reading it now it sounds like Mel read the documentation and knew that an overflow was a jump condition which he could use to eliminate the test from the loop. Or the code was targeting the LGP-30 (see linked comment on the orange site.)

That fake scrollbar makes it unreadably awkward on mobile. If I remember I’ll check this out on desktop tomorrow

The Mel’s story is interesting by itself due to huge difference on how programmers worked and thought when dinosaurs in suits were smoking in the airplanes and drinking whiskey in the offices. But I also like how much better the story becomes when comments are added, also supported with biograph and critics.

This kind of reading support generated automatically would be perhaps a good way to use LLMs BTW.