SLOP8350: Generation Loss
Copies leave clues. In this course, you will investigate what changes when recordings, images and texts are copied, reconstruct possible histories from incomplete evidence, and decide what is worth preserving.
In the first workshop you will put a page through a photocopier ten times and watch it become unreadable. Beside it sits a digital copy of the same page, made ten times over, byte-for-byte identical to the original.
Generation loss is not inevitable. That is an awkward thing for a course called Generation Loss to admit in week one, and it is where the real question starts: which processes change a copy, what traces those changes leave, and when a trace is good enough evidence to say where something came from.
What you will do
Every Wednesday you work on a specimen. You measure what changed, you argue about what the change proves, and you deposit the result in theClass Copy Archive for a later week to use.
By week six you will reconstruct a family tree of hand-copied texts from the mistakes they share, and name the version nobody in the room has seen — then find out where you were wrong. After that the course tests the method until it breaks: on a replication model, on error-correcting codes, on a restoration that invents detail, and on a provenance record with a hole in it.
Scribal error, perceptual audio coding and recursive resampling are three different tests of the same method, not three examples of one mechanism. The course is careful about that difference, and so is the marking.
Who it is for
Postgraduate students interested in archives, media and evidence. You need to be comfortable reading an argument and using a spreadsheet. No Latin, no virology, no programming and no information theory are assumed — each is introduced where the course needs it, with worked numbers.
A laptop and the files on these pages are all you need. Nothing in the core course requires equipment, paid software, a cloud account or code you write yourself.
How it runs
One-hour lecture Mondays, two-hour workshop Wednesdays, twelve weeks with a break after week six. About eight hours a week including preparation and assessment work. Four assessments: an investigation you design, a reconstruction from evidence, a preservation argument you have to cost, and a five-minute evidence question at the start of every workshop.
Every assessment here can be answered with the evidence does not settle this, where the evidence genuinely does not settle it. Saying so well is worth full marks.
The timetable
Every week pairs a Monday lecture with a Wednesday workshop and a five-minute Generation Check at its start. An assessment's own due date can land after the week it was introduced — worked practice keeps going in the weeks between — so the table below places each one on the week it actually falls due.
- Week 124 February 2027
The office that ate itself
- Week 23 March 2027
Hiss as a signature
- Week 310 March 2027
What the ear cannot hear
- Week 417 March 2027
The poor image
- Week 524 March 2027
- Week 631 March 2027
Reading the damage backwards
- Week 714 April 2027
When ancestry becomes hard to read
- Week 821 April 2027
Copying without loss
- Week 928 April 2027
The curse of recursion
- The Stemma due 30 April 2027
- Week 105 May 2027
- Week 1112 May 2027
Chain of custody
- Week 1219 May 2027
The original that never was
- The Lossless Argument due 7 June 2027