It employed a Hamming Code to protect against lightning induced drop outs. This involved taking two words a chosen interval apart, adding them into a check-sum and recording that check-sum down stream by the same interval. As a consequence one could punch holes in the tape about the size of a paper punch and not loose any information. The problem was that a tape splice edit produced a giant pop.
I spotted a TRW 16 bit multiply-accumulator that could operate with a 110 nano-second period. Back-of-the-envelope calculations showed that one such device was sufficient to perform more than 32 channels worth of such operations in the 20 micr-second period of one 16 bit sample period. "We" got a contract to develop such an editor from 3M and delivered a working system in about one year.
I hired one hardware and three software engineers on a part time consulting basis. The programmers all worked for a local microprocessor based manufacturer of telephone transmission test equipment. My hardware guy was a MIT EE who worked for aerospace. This was an evenings and week-ends project and all kept their day jobs. This was, IIRCC, 1978, and we paid our consultants $35/hr. It was exhausting for them but they more than doubled their income for that year. Me, not so much. I should have demanded their deal!
We were using Motorola 6800 microprocessors and I had partitioned the system into four separate, interacting processor systems. One system consisted of the edit hardware, which executed digital cross-fades using stored PROM coefficient tables, MSI logic and the TRW chip, another was the tape machine controller, another was a SMPTE time code reader and sync machine that could synchronize two machines for assembly edits and the fourth was the user interface machine.
The actual cost of the whole project was around $500,000, including hardware. 3M eventually supplied us with recorders on which to test our system. I don't know what "we" charged 3M. I was "just" the project manager. I understood the application and conceived the over all approach and then got burned for my efforts. If "we" charged 3M $1,000,000 at that time it would not have been too outrageous.
In today's environment there are probably lots of guys and gals who would be quite happy to make $80,000 for a year's work and who have the knowledge but not the day job. Make that $100,000, including paid medical of the caliber of Kaiser and one would have choices. One or two hardware guys, depending on whether one engineer could adequately handle both the digital hardware and the analog function modules, two or three programmers and overhead and we would be under $2,000,000/year for development of the full blown "professional" system. A secondary focus on a consumer type game could produce revenue within a year or so.
I have always liked blue skys.
As the Dutch said while fighting the Spanish: "It is not necessary to have hope in order to persevere."
You'd need some start up cash - probably $500,000 or so - but you certainly wouldn't need $50m.
Second Life charges outrageous land rents - a lot of people are paying >$200/mth for a not very powerful virtual server.
I don't think cash is the issue. Nor are processor cycles - clouding should give you all the cycles you want. It might not give them instantly, but some lag wouldn't necessarily be a huge problem.
