I will attempt to gather something. I bet Lothar over at HRA has good info. My guess is the labels will not be as helpful, but it’s worth asking. As I said above, others are working on this problem with the blockchain. I know one guy in a previous WeWork (Benji Rogers of PledgeMusic) who was doing this from a digital licensing pov. It seems that went nowhere, but he might have insights.
Give me some time.
You know everyone as well… Any ideas?
I too am genuinely curious about the current state of affairs as well, but I am worried about the future too.
Another option: If the overall concern for MQA is the baggage, maybe it’s time to fix the baggage. But let’s talk more provenance before we go down that rabbit hole.
PS. Yes, you can have my findings… I’m sure you will do good with it.
One possibility that would frustrate me greatly is if the labels keep this under wraps for fear of losing revenue and somehow sliding back to the Napster days. Foolish I think, but one never knows.
We must also make sure we don’t confuse the definition of provenance, how it has frequently been used with how MQA is using it. We’ve always used it to gauge the honesty of the labels. For example, did they take a 44.1 track and just upsample it to 96k and 192k. This happened several times.
MQA’s use of the term involves zero honesty checks and balances at the labels however. If the label says a 44.1 recording upsampled to 96k then signed-off on is authentic, then we have a “real” fake.
Anyway, I will ask around to find what I can about this topic. It would be great if the MQA team provided some info or even a contact at a label to discuss the issue. Would be a very interesting topic I’d love to write about.
Since we seem to agree the “A” part can be solved (by adding an optional hash (or better, signing key) field to open formats), one thing that neither this nor MQA solves is “trust”.
I say this as someone who wholeheartedly thinks displaying the signatures is well worth the effort you’re putting into it, and is a great idea, but how do I trust a label who “authentified” a file as coming from the artist, when it in reality went through a batch processor, when the purveyor of said authentification has proven deeply disingenuous and lacking in transparency (see the whole “the format is lossless” arguments) ?
In other terms, how do you trust not one, but two untrusted providers ?
Maybe. Or, if I hate Grundman’s work, I’d stay away from it. Although I have no inside information, given the volume of released MQA, I suspect the majority of authentication is going to be label employee. As in, “Yes, MQA processed the files as given to them by us.”
Given his knowledge of tech vs most other musicians, I would have though Peter Gabriel would have been all over this. Imogen Heap though seems to have taken up the tech banner.
Is it though? Dolby was the worst thing to happen to video and I have no desire for MQA to replicate that in music, no matter the extra benefits.
@Carl’s statement while correct in that context, can not be used verbatim in other circumstances.
Roon decodes the files, so obviously it knows more than just the “container”. It knows exactly what data it has.
You can easily infer much about the upper bound of what the content can be from the data in its container, but not so with the lower bound.
There is a difference between knowing whether it’s an oil painting or watercolor painting, vs know if it’s a Danny Dulai or Picaso masterpiece.
Notably, it doesn’t even always make sense to know if the file has been upsampled because the artist who created the content may have used 44.1/16 samples for the drums but a microphone with a 96/24 ADC. When the final recording is made at 96/24, is it real or is it upsampled?
So back on the example you commented on, an AAC file that is containing lossy data can not possibly be high quality because the container can not hold lossless data. Whatever comes out of the AAC has to be a degraded form of whatever went in into it.
“Authentication” is just another example of how MQA marketing speak re-defines terms commonly in use, gives them another meaning, and thus deceives the public about what they are doing. There are many examples. The biggest was originally referring to MQA files as “lossless” (they aren’t), and since that deception fell apart they call the files “perceptually lossless” or imply that they are actually lossless, without specifically using the word “lossless”.
Authentication can be by any record company flunkie; how do you think they are “authenticating” thousands of albums in a short period of time?
We know for a fact of albums that were turned into MQA where none of the artists or recording principals has anything to do with the process, as the relevant people have said so. And is some cases they weren’t happy with the sonic results.
BTW, who is “authenticating” 50 year old or more albums where all the principals are dead?
If they don’t tell us the details of the “provenance” of a master source, the “authentication” means zip.
And Qobuz in no way endorses MQA. They are providing “MQA-CD” files only b/c 2L has decided only to provide those and not standard 16/44.1 files. Just another way that MQA is deceptive, as “MQA-CD” is a misnomer - the files don’t meet the Redboodk 16/44.1 standard.
Yes, it’s still off topic when discussing provenance.
Every MQA topic on every forum turns to a shitfight because everyone mixes all the topics together.
By isolating provenance out, I’m hoping to have a reasonable discussion on the importance of provenance, and how MQA may or may not solve the problem (if there is one).
At this point, it’s clear MQA does not solve it yet because it lacks a system of conducting verification by the public, and the ability to present that – but it’s possible to fix that.
So, why even bring up MQA? Well, who else is closer to getting into the data ingestion pipelines of all the record labels of the world?
Ignore MQA’s marketing then, and focus on the idea of provenance.
This has been pointed out above multiple times, and I’ve proposed that the problem itself goes away over time. See above about big labels vs Bernie Grundman. You can always choose to say “signed by big label XXX is a useless signature” – imagine if Roon could do that automatically.
So then the signatures will suck and be ignored.
Good question. Who do you trust? Is there someone who is better than everyone and anyone?
The alternative is what we have now: a random person who stays anonymous. I’d rather know it was random person at a big label vs someone I can trust.
100% agreed. Presentation and public verification is required for any provenance system to work.
In a case like this, even though it’s a bit of a philosophical bait for the reactionary (in the sense it’d take a certain type of mindset to seriously argue that Stan is a Dido song, or that hip-hop isn’t music), wouldn’t the already available computational analysis (which was referred to in the previous thread, IIRC, there’s a few others, I remember stumbling on a command-line tool that recognized mp3 frames) fall flat on its face when you’re compounding the different sources ?
(here again, user-facing analysis display would be a rather cool feature to add)
The problem is it’s difficult to abstract the many distortions of Bob Stuart if what you want is to pivot MQA, Ltd to being a trusted information provider.
So talk provenance alone. Get past your bias and focus on the first part of this topic. It is possible to own an electric vehicle and hate Elon Musk if you get past the bias.
Agreed, but would it be possible to take your hands off the wheel if Elon Musk had been proven to being mendacious in his claims to deliver a safe self-driving experience ? Because that’s what you’re asking.
Yes, I don’t weigh in personalities or marketing much if the science is right.
Also, most importantly, it wouldn’t stop me from discussing the pros and cons of electric vehicles and what needs to be done to make the solution right. I wouldn’t feel the need to do Elon musk bashing and would be able to focus on the greater topic.
The problem I have with the chain you’re proposing - hence my previous question on trust - is it appears flawed throughout, both on the label side of things and on the information distributor’s side. I mean, I’m sure we’d agree that it’s possible to trust both MQA Ltd’s key provider, and the math behind the signing process, but given that it’s apparently possible to truncate the files and still have the authentication light come on, I personally don’t see how any of it is fixable, given that neither the implementation (i.e, in your car analogy, the science) nor the provider (and @Archimago touched on this more eloquently than I could here) would appear to be reliable in this case.
I’m not suggesting to throw away the provenance baby with the bathwater here, but just wondering aloud what’s fixable, and how, and what isn’t.
It’s not a argument about Bob Stuart as a person; it’s point about the trustworthiness of MQA as a company. They’ve repeatedly shown a penchant for what can be called deception and obfuscation, or even lying. It’s therefore difficult to trust anything they say.
This isn’t an ad hominem argument, but goes to the corporate culture.
For instance, Roon is fairly easy to trust, as you’ve shown yourselves to be reliable with the facts. So if something is said to me by Roon I tend to trust it; if it’s said to me by MQA I tend to try to analyze the statement to see where they are trying to get away with shading the truth. I wouldn’t take anything they say at face value.
I wouldn’t want to trust MQA or anyone to provenance. A good solution here required a system that is very hard to defeat by any group of players, including the provider of the system. You MUST consider MQA Ltd a malicious entity if they are the provider of the secure system.
It’s unclear if this achievable when the cabal could be the 3 major labels… This is similar to the problem of CAs of HTTPS… China can break that entire system by having a mandatory “bad CA”.
Remember, MQA does not currently provide viable provenance for end users. That was determined early in this topic. The discussion here is to find out what would that solution look like.
MQA has baggage clearly so it’d be nice to do this with a fresh start without MQA, but I don’t see anyone attempting to solve the “impossible problem” that Bob Stuart is tackling with the labels: changing the data ingestion pipeline.
The problem with the MQA “authentication” is that essentially they are saying “trust us and trust the record label”. Really?
What we can agree on is that provenance should mean the provider tells us the history of the file, MQA or not. This information is almost never available, unless we are talking about the small specialty labels. And even then it often has to be specifically required, it isn’t provided as a matter of course.
I am hearing artists complain that all the labels aren’t set up to receive MQA directly from artists. This came up at AES in NYC just this last month. That tells me signing will come from artists if we demand it from the labels. No one wants Joe Shmoe at big label’s signature.
