By profession I am a chemist and used to work for the UK Atomic Energy Authority ( but am now retired and program for fun). I learned my computer skills there, growing up with the early computers. The first I had access to was an IBM 360 mainframe and we had to submit our programs to it as batch jobs; the programming being done in Fortran 4 and the archaic about it was that we had to punch our code onto Hollerith cards and send the card deck to the central computer for processing. With with luck we would get one or maybe two runs a day. In addition there were the inevitable typing errors and we would have to replace any faulty cards and submit a new card deck. Getting even a simple program to run would take a day or so and larger programs comprising several hundred cards could take weeks to debug !
Inexorably, technology advanced and we progressed to using teletype terminals connected via a serial link to the mainframe computer. The baud rate of this link was a pathetic 110 baud, but we were delighted that we could now sit in our offices and key in the programs from there; the actual execution of the program was however still under batch control and we still had to wait for the line printer output to discover how successful we had been. Then technology moved on and we moved a little faster using remote-access dumb VDU terminals in text only mode at a baud rate of 1200, but jobs still had to be submitted batch-wise and there was as yet no graphics display. Then after a year or so there was an upgrade to IBM's TSO (Time Sharing Option) which allowed us to at last have full interactive computing in real time; there would be several hundred users all possibly logged on at the same time and we would all have multiplexed access to to the mainframe computer which would sometime crash out if there was too much demand on the system. At the same time there was a limited number of graphics terminals now becoming available, but the baud rate was limited to 9600.
Then in the late 70's there was a requirement for real time control of scientific experiments and for the automatic collection of data and several mini computers with suitable analogue to digital interfaces were acquired . These naturally had to have controlling software and no commercial product was then available; it had to be written. The minicomputer was a Perkin Elmer 16 bit system, with 64k bytes of memory and half of that was taken up by the operating system. Drastic measures were called for if we were to succeed. Writing in Fortran was infeasible because of the size of the resultant program and the only viable solution was to write all the software in assembly and also remove much of the normal operating system components, such as its command processor and file manager. We in effect built our own expert system and with it managed to control experiments which produced several hundred analogue signals for logging and interrogation by users. Real time control of experimental parameters such as temperature, pressure and gas composition was also provided as was a communications link so that users could interrogate and control their experiments in a reactor while situated a mile away in their home building - yes it was a large site.
Then a year or so after it had all been completed we had the launch of the IBM PC (1982) and experimenters started to look at them as a more economical method for data acquisition and control. Yes, it really made a tremendous impact. The mini computer system I referred to cost in the region of £100k and comprised several six-foot instrument racks. Then only a few years later we find that a PC with some installed data acquisition cards can be bought for an order of magnitude less, complete with commercial software and can sit on ones desktop ! This was time to stop fighting the tide and with the cut back in government support for research into commercial reactors, there was an opportunity for staff to retire early, which I took and now can spend my spare time indulging my interest in the fast growing world of computing. There is always something new too learn and its a lot more fun than doing crosswords.