S.PAE@DEEP-THOUGHT.MIT.EDU.UUCP (03/11/87)
> From my experience with error correcting modems, notification of error > correction would be pointless. > ... > If an error correcting scheme can promise me no more than one uncorrected > error every six months of use, I'll be happy. Nothing is perfect, and it is > dangerous to make things almost perfect, because it leads to humans putting > too much faith in such systems (It has to be right, the computer says so.). MNP modems were discussed in TELECOM around 6 months ago. This family of protocols packetize the data, using the CRC-16 polynomial for error detection. DEC used the CRC-16 polynomial for its DDCMP protocol. If used with appropriately small packets (I believe less than 4K bits), CRC-16 will detect all but a miniscule (something around 1 in 10**14) number of the errors. The main problems with MNP are: 1) (At least at 2400bps) the latency caused by the packetization and timeouts makes the character echo time a significant fraction of a second. Interactive use is, at best, annoying. 2) There is no protection for data corruption between the modems and the computers at the ends of the connection. To assure reliable data transfer, there must be a secondary data-integrity mechanism in place. Since MNP can't, by its design, do the whole job, why not use the secondary mechanism exclusively? For detailed information about what I've summarized above, please consult the TELECOM archives. At my job, we have the error correcting turned off on all of our modems. I'll occasionally turn it on when I want to monitor something from home for many hours (and don't have a lot of typing to do!). It goes back off immediately afterwards. --phil