Re: 30% chance of failure?
and entirely didn't bother with error correction, so even the original cassette was hard to load.
Nor did the original Speccy format. It just recorded an additional byte (or maybe just a bit) at the end of the data stream which when you counted the number of set bits made the total odd or even or somesuch. That's why you never found out about a bad load until right at the very end. At least the beeb format split the data into blocks and checked each block.
The original speccy format was 'reet crude'. When recording it toggles the output signal on/off. The interval between each transition indicates if it's a 1 or a 0. After eight transitions you've got a byte. After 8192 transitions you have 1kB. No blocking, no error correction. Just a stream of bits.