Typesetters in the '80s - Computerphile

Key Takeaways

perhaps one shouldn't be putting a date stamp on these videos but since it's early 2014 I'll say happy New Year everyone I'm absolutely delighted that so many of you uh enjoyed so much the previous video we put out which of course was the great 202 jailbreak of what I want to do today is to carry on that story given that you all seem to enjoy it so much first of all to explain how we in Nottingham got involved in the whole 202 Saga but then more importantly to lead you on from that era to a time only about two years after that when it did become possible for everybody to do their own type setting so if we set the time clock now back to about mid 1982 which is about just over two years after the great jailbreak Saga we in Nottingham had heard all about what Bel are done although we didn't know about the jailbreak all that we knew was that they were using a 202 typ Setter and we wanted at the time to typ set our own mathematics papers and you might say well why would you want to do that uh can't you just do them on a typewriter well of course the special symbols in math cause a problem they don't look good at all sometimes the math symbols had to be Inked in by hand and uh we had uh an an absolute perfectionist in the Mass Department called George George pester and he had been brought up in Oxford University near the Oxford University press and he wanted all our exam papers to look like works of art produced by o this costs money this costs serious money particularly if it's mathematics papers which were at the time the bulk of uh what we wanted to do it was costing the university in 1980s money $18,000 pound a year to send the type setting to outside firms because we didn't have our own type Setter so wouldn't it make sense to get one of our own well yes George was very persuasive next thing I knew I had a telephone call from the University's registar saying would I lead the team to implement a typ Setter a suitable type Setter uh attached to a uh computer running this new fangle thing called Unix because the university had been assured that we could type set our own papers and that this would in the medium term save us money so thus began the Great Adventure whereas Brian had a team of three I had a team of six and we're talking about late 1982 when we got started now those of you who followed the previous video and followed the links out to the backup documentation may recall that at the very end of the Bell laabs 202 Saga Morgan tler in the states had come to Brian and said look you know there's a new generation of Machinery coming along what perhaps we really should have done all along is pointed you towards our brand new Omnitech in Nottingham we were visited by linotype Paul which was the UK arm of mgan tala they were very honest with us they told us all about the bad boys at B Labs who' got into their 202 and hoped that we weren't going to do anything similar to that and said well you know what we said to them at the end and would say it to you now is maybe you ought to go with this this is the future it's the Omnitech effectively what it was was a high-priced high resolution laser printer it was the right decision in principle but the timing was just awful it was very early on in the uh if you like development cycle of Laser Printers and being a types set manufacturer and not just if you like a lowquality laser printer manufacturer like imagin or Cannon or whatever who were quite happy with 300 dots per inch 300 DPI with Laser Printers is pretty straightforward to get to work relatively speaking but linotype mgala had standards they wanted at least 700 DPI well it was doable in the day but only just just look at the quality of the uh of the catalog and pamphlet alone oh and a keyboard with things to keep a proper types set person happy like thin spaces M spaces stuff like that liner type were intent that this should be a proper typ Setter it was to be high resolution reluctantly they had to agree that the 972 DPI of the um 202 which of course involved smelly developer bromide and all that well if you move to paper you not going to be able to do 900 DPI but they did think they'd found a way to do 720 DPI and yes they had but there were problems at its very best when it was working it could produce very pleasing looking output but the trouble with trying to do 720 DPI is that you need very finely divided and suspended toner inks in liquid also special zinc oxide coated paper everything was fine when it worked well in fact I've got some samples here of the kind of output quality you could get but wo be TI you if you ever skimped on cleaning the thing every night and every morning because what would happen was that the finally divided toner ink would then clog up the tubes that fed the Imaging drum at best it might blow a fuse and then if you replace the fuse you'd find that suddenly the pump would start splitting a tube and pumping toner ink all over theor floor it really was a complete Nightmare and in fact it was not just us the thing was actually withdrawn I think in early 1984 the other big problem with it was that it was the future but it was of course slow notice down here what processes were being used Intel 88 Xylo Z8 and I think it's a signetics processor the hx300 but we're talking here essentially about the dawn of 8bit microchips they were the future but they were not fast enough it very soon became obvious that a faster 16 bit system was the minimum that you needed to run a machine like this so after something like nine months of effort we did succeed in typ setting the mathematics papers for 1983 using the Omnitech but we realized that we just were not going to be able to persist with this inter 1984 it was far too unreliable so in a strange sort of reversal of chronology we decided that what we'd have to do and and line of type UK supported us fully in this decision was to go back a generation and revert to using a good old 202 which we knew that b Labs successfully used and yeah so we took delivery of our linotype 202 Brian and I had been in cor respondence about the fact that we were doing type setting with the Omnitech and it was arguably an even bigger disaster than their 202 was it was at that stage I think that Brian sent me a copy of the vacation memo but of course we both knew that there was no way that we at Nottingham were going to uh use the Bell Labs software uh for a start they couldn't have wouldn't have supplied it to us and also I think I felt by that time that the 202 had had four more years of development work done on it uh I think the 202 was released in 1978 Brian's model of course was 1979 they hadn't really had the time to fix all of the hardware and software problems the low-level liner type software to drive the 202 was perfectly serviceable it was called binary bite and that's what we used in the early releases when Brian was using it it was so full of bugs it was unusable and that was why in the jailbreak so they just you know took over the entire machine hardware and software cleared out all the liner type software uh because they didn't think it was reliable enough and when you face also that the unreliability in the software were interacting with unreliability in the hardware you could see why they went for complete replacement of everything but to L of Ty's great credit the 202 was a great design and by the time it had settled down it gave us no trouble at all yes it was a pain having to process bromine but frankly we just sailed through 1984 type setting burden with um the 202 the only sad thing for me as as a leader of the team and if any of you want to know more about the team we'll put a uh a web link out on this to the paper that web pointers for those of you who like that phrase yeah we'll put a web pointer a web link out to the paper that describes the naum involvement you can find out the names of all the people who helped me so much at that time and I think you might at the end of that get a feeling of complete wistfulness that we had to give this thing up there it was in our Mass basement being commissioned in late 1983 the typ setting season for exam papers started typically in January February of every year so by early 1984 it had to be delivered out of our basement up the hill as as we called it to the exams Department which was a subset of the registar department and it was installed there I sent two of my guys Julian and William you'll see them referred to in the uh cast list of this paper and uh they looked after it interfaced it to Unix and brought back you know wistful reports to me of oh when were we going to get one of our own and so on in fact but the thing into context I've got hold of here a reproduction of our party piece that we used to do for people on the 202 can you imagine that in beautiful fresh gleaming bromide what had happened was that another colleague on the implementation team douglas douglas Woodall had been off to a pure mathematics Symposium I think in New Orleans or somewhere in the USA and he came back he said I've just seen this most wonderful wonderful wonderful t-shirt and um it's got Max equations written out in full and this is what this uh poster if you like is summarizing those of you who are well up in University level physics will know that Maxwell's electromagnetic equations are normally abbreviated using Vector operators called grad div and curl and you can compress it all down into about three lines this is what happens if you expand those grads divs and curls out into full-blown partial differential notation so there we are our poster we'll put a link out to that as well lucky viewers and um yeah it was wonderful but when were we ever going to be able to do that again for ourselves given that our only hope had vanished up the uh up the hill and that we for the next 18 months of of of typ setting winter if you like had to revert back to this which is what I've shown you before pretty good quality dot matrix output but nothing to compete with what would happen with proper fonts and so on in this era you mustn't imagine that the 202 and its characters were able to do splines and arcs makes very clear this difficulty with getting characters to look good on a coarse resolution