Karl Miller wrote:
> On Mon, 27 Mar 2006, Mike Richter wrote:
> 
> 
>>2. Manufacturers of recordable media guesstimated lifetime in centuries,
>>not decades.
> 
> 
> I can only relate my recollection that suppliers of CDROM products would
> limit lifetime to ten years.
> 
> 
>>3. English text - ASCII, ANSI or otherwise represented - is highly
>>tolerant of error. Executable code is not. Audio is moderately tolerant
>>for most listeners in most cases.
> 
> 
> This I do not understand. With each bit being significant for text, how
> can text be less subject to error?

When the claims were being made, lifetime was estimated to be 100 years 
with premium media talking about 200 years. Those numbers were as 
realistic as the recent PC publication claiming that burned media would 
die in two to five years.

Redundancy in English text is very high, usually sufficient for the 
original to be reconstructed when every third character is lost. For 
example:
Th- q-ic- b-ow- f-x -um-ed-ov-r -he-la-y -og

At 10% lost, one gets something like
C-me gaze -ith me upo- this dome-of many-co-ored glass

Finding the original poem given that clobbered first line would not be 
difficult - assuming that cummings's poetry was deemed fit for the 
collection.

Do *not* try that with a computer program. <G>

Mike
-- 
[log in to unmask]
http://www.mrichter.com/