On the Reverse Engineering of Big Old Machines

Intellectual Habits

image link-topic-sf0.jpg

Whatever one does to an antique machine, one has got to forget one's own personality and put oneself into the mind of the maker ... Enter the mind of the maker and do it his way. - George Daniels. [Interviewed in 2003].
I think part of the key to making accurate 3D models is fully understanding the process that went into making them. Simply trying to emulate the shape directly can lead to a complex inaccurate model. - Daryl Bender. [Personal communication, 2015].

1. On Becoming an Engineer of 1886

Do you recall the part of The Hitchhiker's Guide to the Galaxy series in which Arthur Dent tries to teach the computer how to make a proper cup of tea? (If you don't - if you've never heard of this - then stop right now and read these books by Douglas Adams. The passage in question here appears in The Restaurant at the End of the Universe.) Arthur Dent doesn't just tell the "Nutri-matic" computer about the boiling point of water.

" He told the Nutri-Matic about India, he told it about China, he told it about Ceylon. He told it about broad leaves drying in the sun. He told it about silver teapots. He told it about summer afternoons on the lawn. He told it about putting in the milk before the tea so it wouldn't get scalded. He even told it (briefly) about the history of the East India Company."

Real reverse engineering is like this. It isn't sufficient just to pop a part in a 3-D scanner and print a new one. That just gives you a copy of a worn part. It isn't sufficient to engineer a new part to 21st century design practices, even though your new part works. That just gives you a working machine which was obsolete before your grandparents were born. For this to be worthwhile, the machine must become new again. If the project of reverse engineering an old machine is to have any meaning at all, you must re-create the part just as the original engineer designed it. If your machine dates to 1886, you must learn to think like an engineer of 1886; you must become an engineer of 1886 and build the old machine as new, once more.

Of course, it is wise to remember that Arthur Dent's attempt to teach a computer to make a real cup of tea nearly resulted in his death.

See also Makers' History.

2. Documenting Why You Think You Know Something

It may seem counterintuitive, but simple, direct, practical knowledge is the bane of true understanding. It doesn't matter in the least that you know something, because you may well be wrong. What matters, really, is that you know how and why you think you know something. Only then can you hope to correct your errors.

Documenting why you think you know something is the only way to pass on knowledge. You cannot just pass on "the plain and simple facts," because these are at all times no more than your own always-questionable system of beliefs. Pass on your reasons and sources for knowing, so that your heirs can correct your mistakes when you cannot.

3. Admit that You Don't Know

I have over the course of my life been acquainted with a number of very intelligent people who were unable - entirely unable at a basic level - ever to admit that they did not know the truth. Sometimes my involvement with these people was extensive and of considerable duration.

Looking back over the decades, it strikes me that time and again their certainty was misplaced. They knew they were right, but they were wrong. But in no case could they ever admit this - it wasn't vanity, but rather a psychological inability to understand that their views were other than the eternal truth of the world. In some cases they simply held on to old beliefs after they had been quite clearly shown to be wrong. (Sometimes these beliefs were in larger spheres - public affairs and such - but at other times they were of a quite technical nature where ideas were falsifiable in the Popperian sense.) In other cases - subtler, but more pernicious - they shifted to a new set of beliefs and forgot instantly that they had ever believed otherwise.

Knowing that something is true - which in scientific thinking is impossible - is a great enemy of understanding. In reverse engineering, this false certainty must be guarded against by the careful documentation of what really is there in the machine - and what is not.

For an example, consider the trimming knives of the Thompson Type-Caster. I am unaware of any surviving, labeled, new-old-stock packets which confirm that particular examples are original production. I am unaware of any surviving original Thompson Type Machine Company or Lanston Monotype Machine Company engineering drawings by which the machine's engineers could convey their design intention. Surviving examples - any of which or none of which might or might not be authentic - contain sufficient variation that they would not normally be described as the same engineering revision level of the same component. In short, we do not know with certainty exactly what a 79TC2 or 80TC8 Thompson Trimming Knife looks like - even though several items fulfilling their function are in daily operation today.

In reverse-engineering the Thompson, we must first document what these differences are. Then, to fulfil the need of creating new parts, we must create a plausible compatible new design based on an understanding of this documented existing practice. In doing so, we must identify this design as new, so as not to confuse further the issue by adding to the collection of anonymous existing practice.


Select Resolution: 0 [other resolutions temporarily disabled due to lack of disk space]