As someone who has also tried to move vinyl to digital I can say firmly without a doubt that vinyl is an awful medium for fidelity preservation. As @mike70 said it is a mechanical system and it's a remarkably finicky when you try to wring out the full capabilities of it. They were never made to work out to their technical limits regularly because the vast majority of people just wanted to listen to their music and would accept a simulacra of a performance if it was cheap and easy enough to use.
One of the main thing people don't understand in the analog/digital divide is that there is a difference between an electro-mechanical system that interfaces mechanically (and thus destructively) and naively presents the electrical interpretation for direct amplification to an opto-computational system that interfaces non-destructively, encodes data for structured interpretation, and can be modified before amplification.
(Grant me the freedom not to quibble about specialist equipment in the next part)
Every time you play a record you slightly destroy it.
The needle will wear against the record groove and the groove against the needle. The change will be small but it will eventually change the actual signal because the wear will not be perfectly even. This goes quadruple for an improperly balanced system, the needle will apply a mechanical bias in some direction that will in the worst case will lead to it eventually just throwing the needle out of the groove. The sound will decay before it gets to that point, but again the point of a record is for casual playing and it will accept a surprising amount of obvious content degradation before it fully fails. Magnetic physical media (e.g. analog tape and the mechanical aspect of hard drives) requires the same rough concepts of physical control and tolerances but at least it doesn't actively destroy both the content and capability to retrieve the content
The content is presented naively for real time signal transposition.
So assuming you are getting the intended content, now you have a physical range to interpret into electrical values. But there is no intentional computational pause in this process so if something goes momentarily wrong there is no way to attempt to correct for it. This is both a blessing and a curse, for sufficiently small problems the brain will generally be able to compensate without too much attention drawn to the error but since you're stuck with a single attempt it means as your data size grows your error rate grows proportionally. Again, for normal human consumption of music it doesn't matter, but if you're aiming for perfect replication or using pointlessly bloated data files you'd want to stay away from true real time transfer without a buffer.
And even if you had magical buffer that would physically re-check imperceptibly fast (awfully hard if you are limited to 33rpm) you would have no way of saying that the signal found is what was intended. The values aren't quantized and packaged with error correction, so you can't say that value X was supposed to be value Y because of another authoritative structure Z to reference. Also if you did already have an authoritative copy to check against, why don't you just play that?
Low headroom for signal modification
Again, a blessing and a curse. The physical nature of the record meant that you just physically couldn't record all possible signal transitions in the physical range (otherwise you would throw the needle off); so records had to be recorded in a certain manner and they choose what they subjectively felt was the best way in the limits they had and it could sound pretty damn good when done right. Digital media doesn't have the same kind of limitation for better or worse, so you can push the limits of taste further but it's a magnitude (scaler?) not a vector (e.g. Loudness Wars). Also digital media had to be converted back into analog at some point for consumption and historically DACs were kind of not great unless you spent a lot of money.
And all of this assumes the physical media hasn't been subject to damage (like storing them sideways in even a modest stack), isn't covered in dust, etc. They deform from true flat planes way too easily and it's a real pain to get them flat again. There isn't a physical standard, at least as far as I can tell, that will guarantee best playback on all records without doing research on the company that did that pressing and that particular run.
So yes, vinyl sucks as a medium and if it sound quality you are looking for then a properly digitized file will always beat it out. The problem is the inability to get those digitized files. There is no way to go to a record company and say "Here is proof that I own all of these records already, I will pay you $5 a record for digital copies of what I already own". You can try to digitize your own records but it's very very painful and incredibly slow because you have to do the metadata creation yourself and you can't share the workload among other enthusiasts because each one of you is only legally allowed to transfer your own personal record to a personal file. Simply owning the record does not give someone else the right to give you a digital copy even though it's theoretically the same content for all meaningful purposes.
In a sane world we would just create a dummy record system that would let people fiddle with lots of finely weighted knobs and then spin a 'record' and simulate the needle drop while playing back a bit perfect digital copy made from the original master using laser transcription. Then we'd lock the padded door and let them get on with their lives.
One of the main thing people don't understand in the analog/digital divide is that there is a difference between an electro-mechanical system that interfaces mechanically (and thus destructively) and naively presents the electrical interpretation for direct amplification to an opto-computational system that interfaces non-destructively, encodes data for structured interpretation, and can be modified before amplification.
(Grant me the freedom not to quibble about specialist equipment in the next part)
Every time you play a record you slightly destroy it.
The needle will wear against the record groove and the groove against the needle. The change will be small but it will eventually change the actual signal because the wear will not be perfectly even. This goes quadruple for an improperly balanced system, the needle will apply a mechanical bias in some direction that will in the worst case will lead to it eventually just throwing the needle out of the groove. The sound will decay before it gets to that point, but again the point of a record is for casual playing and it will accept a surprising amount of obvious content degradation before it fully fails. Magnetic physical media (e.g. analog tape and the mechanical aspect of hard drives) requires the same rough concepts of physical control and tolerances but at least it doesn't actively destroy both the content and capability to retrieve the content
The content is presented naively for real time signal transposition.
So assuming you are getting the intended content, now you have a physical range to interpret into electrical values. But there is no intentional computational pause in this process so if something goes momentarily wrong there is no way to attempt to correct for it. This is both a blessing and a curse, for sufficiently small problems the brain will generally be able to compensate without too much attention drawn to the error but since you're stuck with a single attempt it means as your data size grows your error rate grows proportionally. Again, for normal human consumption of music it doesn't matter, but if you're aiming for perfect replication or using pointlessly bloated data files you'd want to stay away from true real time transfer without a buffer.
And even if you had magical buffer that would physically re-check imperceptibly fast (awfully hard if you are limited to 33rpm) you would have no way of saying that the signal found is what was intended. The values aren't quantized and packaged with error correction, so you can't say that value X was supposed to be value Y because of another authoritative structure Z to reference. Also if you did already have an authoritative copy to check against, why don't you just play that?
Low headroom for signal modification
Again, a blessing and a curse. The physical nature of the record meant that you just physically couldn't record all possible signal transitions in the physical range (otherwise you would throw the needle off); so records had to be recorded in a certain manner and they choose what they subjectively felt was the best way in the limits they had and it could sound pretty damn good when done right. Digital media doesn't have the same kind of limitation for better or worse, so you can push the limits of taste further but it's a magnitude (scaler?) not a vector (e.g. Loudness Wars). Also digital media had to be converted back into analog at some point for consumption and historically DACs were kind of not great unless you spent a lot of money.
And all of this assumes the physical media hasn't been subject to damage (like storing them sideways in even a modest stack), isn't covered in dust, etc. They deform from true flat planes way too easily and it's a real pain to get them flat again. There isn't a physical standard, at least as far as I can tell, that will guarantee best playback on all records without doing research on the company that did that pressing and that particular run.
So yes, vinyl sucks as a medium and if it sound quality you are looking for then a properly digitized file will always beat it out. The problem is the inability to get those digitized files. There is no way to go to a record company and say "Here is proof that I own all of these records already, I will pay you $5 a record for digital copies of what I already own". You can try to digitize your own records but it's very very painful and incredibly slow because you have to do the metadata creation yourself and you can't share the workload among other enthusiasts because each one of you is only legally allowed to transfer your own personal record to a personal file. Simply owning the record does not give someone else the right to give you a digital copy even though it's theoretically the same content for all meaningful purposes.
In a sane world we would just create a dummy record system that would let people fiddle with lots of finely weighted knobs and then spin a 'record' and simulate the needle drop while playing back a bit perfect digital copy made from the original master using laser transcription. Then we'd lock the padded door and let them get on with their lives.
Last edited:
