6 ms·

Mediachain enivisions a blockchain-based tool for identifying artists’ work

54 points by jrbedard 10y ago

duncancarroll 10y ago
It's a great idea with a lot of utility, but my question is: How does dispute resolution work? I.e. How do you prevent bad actors from beating the original artist to the punch, and registering someone else's artwork as theirs on the blockchain?
- denisnazarov 10y ago
  The short answer is reputation. Consumers of the data should be able to filter by cryptographic signature to prioritize users that have more reputation over ones that don't. A timestamp is just one piece of metadata. For example, on a platform like Soundcloud there are many Rhiannas but it is obvious which account the "real" Rihanna is because she has the most followers, likes, etc. Popularity can be another data point. Projects like Blockstack already offer solutions that take a similar approach, allowing you to link your social media accounts so your identity is more trustable. We also envision developers and trusted organizations like Creative Commons or DPLA creating indexes on top of Mediachain to help users filter the data.
  - dublinben 10y ago
    Reputation is not as easy to measure as simple popularity. There are very many "content creators" like LadBible who are orders of magnitude more popular than the actual content creators they steal from.
  - Avshalom 10y ago
    That's not much of a "how". Okay dispute happens Alice has Reputation X and Bob has Reputation Y. So first who or what does Alice dispute to and second so what: if X > Y, Alice becomes the owner?
denisnazarov 10y ago
Hey HN, I'm Denis, one of the creators of Mediachain. Happy to answer any questions! This Co.Design article does a great job of explaining how Mediachain helps creators: http://www.fastcodesign.com/3060426/could-blockchains-solve-the-webs-image-attribution-problem http://www.fastcodesign.com/3060426/could-blockchains-solve-... Here is our official fundraising announcement with more details: https://blog.mediachain.io/mediachain-labs-funded-by-union-square-ventures-andreessen-horowitz-to-build-a-universal-open-337ee690f61b https://blog.mediachain.io/mediachain-labs-funded-by-union-s...
- paavokoya 10y ago
  Hey Denis, What differentiates your company from the hundreds of other "blockchain-based" identity companies that do almost the same thing? I've seen a lot of companies throwing around these exact buzzwords and fancy landing pages for 5 years now and most just seem to be raking in VC funding while fading into obscurity a year later. There's a list if you scroll down on this link: http://www.blockchaintechnologies.com/blockchain-companies http://www.blockchaintechnologies.com/blockchain-companies
  - denisnazarov 10y ago
    We're creating a completely open, decentralized media library and our goal is to help creators reach their audience directly through their content. We believe that value should follow the content, instead of flowing from a platform. Today, Mediachain helps preserve attribution for creative works wherever they go, so you can know the creator of that anonymous viral image you see in your feed. In the future the same infrastructure will enable financial exchange directly between a creator and their audience through their works. We've written some more about the implications of Mediachain here: https://blog.mediachain.io/the-gif-that-fell-to-earth-eae706c72f1f https://blog.mediachain.io/the-gif-that-fell-to-earth-eae706... https://blog.mediachain.io/what-a-blockchain-for-music-really-means-e2f8dc66d57d#.y4xge49cc https://blog.mediachain.io/what-a-blockchain-for-music-reall...
    - paavokoya 10y ago
luk3thomas 10y ago
What happens if someone crops and scrubs the metadata of the original image and then shares it? Can mediachain identify the original source of the new image?
- denisnazarov 10y ago
  Any metadata that is stored in EXIF or part of the file can be easily stripped out. This is why Mediachain uses perceptual recognition technology similar to Shazam or Google Image search to identify media based on how it looks or sounds, automatically resolving to metadata stored in Mediachain. Near duplicate image detection is quite far along and works very well even for images that are cropped, distorted, etc. More on our approach in this post: https://blog.mediachain.io/perceptual-resolution-9c00ad5ca555?source=latest https://blog.mediachain.io/perceptual-resolution-9c00ad5ca55... Details on the implementation in this RFC: https://github.com/mediachain/mediachain/blob/master/rfc/mediachain-rfc-3.md https://github.com/mediachain/mediachain/blob/master/rfc/med...
CPLX 10y ago
> Imagine being able to connect with the artist of a viral GIF you see in your feed, learn the history or origin of any image, or automatically reward a musician whenever you press play. OK, I've gone and imagined it. I'm not entirely convinced that any appreciable number of people actually want to do these things though. Is this a clever application of a trendy technology creating a solution that's searching for a problem? What are the real life business model use cases for such a tool?
- denisnazarov 10y ago
  Our goal is to make the monetization of media follow the content, instead of flowing from the platform like it does today. Today, Mediachain can automatically link an image to who made it. That same channel can be used to transfer value directly through content in the future. We believe that this will create a huge opportunity for developers to build new media applications that reward creators directly through their content, no matter where it is.
  - chillingeffect 10y ago
    How is it different from, e.g. hashing or fingerprinting an image a la tineye? Does your system require metadata along with the image? If so, what about once an image escapes the metadata?
    - denisnazarov 10y ago
      The perceptual recognition technology is similar to Shazam or Google Image search (or Tineye), identifying media based on how it looks or sounds. This allows Mediachain to always point back to information about an image, even if its metadata is stripped out as it goes viral, for example. Near duplicate image detection is quite far along and works well for images that are cropped, distorted, etc. More on our approach in this post: https://blog.mediachain.io/perceptual-resolution-9c00ad5ca555?source=latest https://blog.mediachain.io/perceptual-resolution-9c00ad5ca55... Details on the implementation in this RFC: https://github.com/mediachain/mediachain/blob/master/rfc/mediachain-rfc-3.md https://github.com/mediachain/mediachain/blob/master/rfc/med...
bmcusick 10y ago
How does your perceptual recognition technology distinguish piracy from Fair Use? Given the impossibility of knowing with certainty which jurisdiction the end user is in, which Copyright scheme are you assuming? Can works be registered as Public Domain or as subject to a permissive license such as Creative Commons? This sounds like you're making an Internet-scale DRM scheme. What precautions are taking to protect people from creators trying to take more than copyright provides?
- denisnazarov 10y ago
  The goal of Mediachain isn't to enforce scarcity of media online, but to help creators benefit from the scale of sharing. Today, an image can easily go viral without viewers knowing who the author was. Mediachain makes sure that the creator can be present wherever the content goes, automatically. Our view is that bits are easily reproduced and there is no way to stop that, and until now there's been no easy way for creators to benefit from that. Mediachain allows creators to connect directly to their audiences through the content itself. This way, gratitude can be exchanged through attribution, and in the future, even a payment directly through the content.
untilHellbanned 10y ago
NYC-based Monegraph anyone?
sandworm101 10y ago
Ok. How about this:http://i.imgur.com/heXkHzU.png http://i.imgur.com/heXkHzU.png It's a partial screenshot of the article in my browser. I intend to use it in a lecture as an example of the complications behind fair use. (irl I really do teach such a lecture). Who is it attributable to? The artist of the drawing, me, techcrunch or mediachain? In short: I am suspicious of any service, and there have been a few, that claims to be able to identify the owner or creator of a work. It's a layering problem, with nearly every digital image being at some level the property of a great many people. (fyi: if this was a law exam, the best answer is "me" in that I created the composition and am well-covered by fair use in my use of it here.)
- parkan 10y ago
  Declaring ownership and provenance with complete certainty is very hard, which is why we don't profess to do it. Our approach is to nondestructively aggregate potentially conflicting claims and allow for more nuanced resolution at read time, as appropriate for a given situation. In your particular example, we would have a chain of derivative works. Obviously, in certain cases (payment routing) a definitive answer is required, but this is already something that PROs and similar organizations deal with on a daily basis, and we can support their arbitration.
  - sandworm101 10y ago
    If you plan on linking money to content, you better have a really definitive answer. That movement of money is essentially the purchase of a license. If a business or individual relies on your answer as to who that license should be purchased from, they do not want to be hit with a copyright complaint should the real owner appear. And if people end up abusing your system to monetize content that they should not, expect all sorts of legal docs in the mail.
    - parkan 10y ago
      Absolutely! Spotify's 30M settlement with NMPA is a great example of this kind of failure in existing systems

54 points by jrbedard 10y ago

It's a great idea with a lot of utility, but my question is: How does dispute resolution work? I.e. How do you prevent bad actors from beating the original artist to the punch, and registering someone else's artwork as theirs on the blockchain?

denisnazarov 10y ago
The short answer is reputation. Consumers of the data should be able to filter by cryptographic signature to prioritize users that have more reputation over ones that don't. A timestamp is just one piece of metadata. For example, on a platform like Soundcloud there are many Rhiannas but it is obvious which account the "real" Rihanna is because she has the most followers, likes, etc. Popularity can be another data point. Projects like Blockstack already offer solutions that take a similar approach, allowing you to link your social media accounts so your identity is more trustable. We also envision developers and trusted organizations like Creative Commons or DPLA creating indexes on top of Mediachain to help users filter the data.
- dublinben 10y ago
  Reputation is not as easy to measure as simple popularity. There are very many "content creators" like LadBible who are orders of magnitude more popular than the actual content creators they steal from.
- Avshalom 10y ago
  That's not much of a "how". Okay dispute happens Alice has Reputation X and Bob has Reputation Y. So first who or what does Alice dispute to and second so what: if X > Y, Alice becomes the owner?

denisnazarov 10y ago

Hey HN, I'm Denis, one of the creators of Mediachain. Happy to answer any questions! This Co.Design article does a great job of explaining how Mediachain helps creators: http://www.fastcodesign.com/3060426/could-blockchains-solve-the-webs-image-attribution-problem http://www.fastcodesign.com/3060426/could-blockchains-solve-... Here is our official fundraising announcement with more details: https://blog.mediachain.io/mediachain-labs-funded-by-union-square-ventures-andreessen-horowitz-to-build-a-universal-open-337ee690f61b https://blog.mediachain.io/mediachain-labs-funded-by-union-s...

paavokoya 10y ago
Hey Denis, What differentiates your company from the hundreds of other "blockchain-based" identity companies that do almost the same thing? I've seen a lot of companies throwing around these exact buzzwords and fancy landing pages for 5 years now and most just seem to be raking in VC funding while fading into obscurity a year later. There's a list if you scroll down on this link: http://www.blockchaintechnologies.com/blockchain-companies http://www.blockchaintechnologies.com/blockchain-companies
- denisnazarov 10y ago
  We're creating a completely open, decentralized media library and our goal is to help creators reach their audience directly through their content. We believe that value should follow the content, instead of flowing from a platform. Today, Mediachain helps preserve attribution for creative works wherever they go, so you can know the creator of that anonymous viral image you see in your feed. In the future the same infrastructure will enable financial exchange directly between a creator and their audience through their works. We've written some more about the implications of Mediachain here: https://blog.mediachain.io/the-gif-that-fell-to-earth-eae706c72f1f https://blog.mediachain.io/the-gif-that-fell-to-earth-eae706... https://blog.mediachain.io/what-a-blockchain-for-music-really-means-e2f8dc66d57d#.y4xge49cc https://blog.mediachain.io/what-a-blockchain-for-music-reall...
  - paavokoya 10y ago

luk3thomas 10y ago

What happens if someone crops and scrubs the metadata of the original image and then shares it? Can mediachain identify the original source of the new image?

denisnazarov 10y ago
Any metadata that is stored in EXIF or part of the file can be easily stripped out. This is why Mediachain uses perceptual recognition technology similar to Shazam or Google Image search to identify media based on how it looks or sounds, automatically resolving to metadata stored in Mediachain. Near duplicate image detection is quite far along and works very well even for images that are cropped, distorted, etc. More on our approach in this post: https://blog.mediachain.io/perceptual-resolution-9c00ad5ca555?source=latest https://blog.mediachain.io/perceptual-resolution-9c00ad5ca55... Details on the implementation in this RFC: https://github.com/mediachain/mediachain/blob/master/rfc/mediachain-rfc-3.md https://github.com/mediachain/mediachain/blob/master/rfc/med...

CPLX 10y ago

> Imagine being able to connect with the artist of a viral GIF you see in your feed, learn the history or origin of any image, or automatically reward a musician whenever you press play. OK, I've gone and imagined it. I'm not entirely convinced that any appreciable number of people actually want to do these things though. Is this a clever application of a trendy technology creating a solution that's searching for a problem? What are the real life business model use cases for such a tool?

denisnazarov 10y ago
Our goal is to make the monetization of media follow the content, instead of flowing from the platform like it does today. Today, Mediachain can automatically link an image to who made it. That same channel can be used to transfer value directly through content in the future. We believe that this will create a huge opportunity for developers to build new media applications that reward creators directly through their content, no matter where it is.
- chillingeffect 10y ago
  How is it different from, e.g. hashing or fingerprinting an image a la tineye? Does your system require metadata along with the image? If so, what about once an image escapes the metadata?
  - denisnazarov 10y ago
    The perceptual recognition technology is similar to Shazam or Google Image search (or Tineye), identifying media based on how it looks or sounds. This allows Mediachain to always point back to information about an image, even if its metadata is stripped out as it goes viral, for example. Near duplicate image detection is quite far along and works well for images that are cropped, distorted, etc. More on our approach in this post: https://blog.mediachain.io/perceptual-resolution-9c00ad5ca555?source=latest https://blog.mediachain.io/perceptual-resolution-9c00ad5ca55... Details on the implementation in this RFC: https://github.com/mediachain/mediachain/blob/master/rfc/mediachain-rfc-3.md https://github.com/mediachain/mediachain/blob/master/rfc/med...

bmcusick 10y ago

How does your perceptual recognition technology distinguish piracy from Fair Use? Given the impossibility of knowing with certainty which jurisdiction the end user is in, which Copyright scheme are you assuming? Can works be registered as Public Domain or as subject to a permissive license such as Creative Commons? This sounds like you're making an Internet-scale DRM scheme. What precautions are taking to protect people from creators trying to take more than copyright provides?

denisnazarov 10y ago
The goal of Mediachain isn't to enforce scarcity of media online, but to help creators benefit from the scale of sharing. Today, an image can easily go viral without viewers knowing who the author was. Mediachain makes sure that the creator can be present wherever the content goes, automatically. Our view is that bits are easily reproduced and there is no way to stop that, and until now there's been no easy way for creators to benefit from that. Mediachain allows creators to connect directly to their audiences through the content itself. This way, gratitude can be exchanged through attribution, and in the future, even a payment directly through the content.

untilHellbanned 10y ago

NYC-based Monegraph anyone?

sandworm101 10y ago

Ok. How about this:http://i.imgur.com/heXkHzU.png http://i.imgur.com/heXkHzU.png It's a partial screenshot of the article in my browser. I intend to use it in a lecture as an example of the complications behind fair use. (irl I really do teach such a lecture). Who is it attributable to? The artist of the drawing, me, techcrunch or mediachain? In short: I am suspicious of any service, and there have been a few, that claims to be able to identify the owner or creator of a work. It's a layering problem, with nearly every digital image being at some level the property of a great many people. (fyi: if this was a law exam, the best answer is "me" in that I created the composition and am well-covered by fair use in my use of it here.)

parkan 10y ago
Declaring ownership and provenance with complete certainty is very hard, which is why we don't profess to do it. Our approach is to nondestructively aggregate potentially conflicting claims and allow for more nuanced resolution at read time, as appropriate for a given situation. In your particular example, we would have a chain of derivative works. Obviously, in certain cases (payment routing) a definitive answer is required, but this is already something that PROs and similar organizations deal with on a daily basis, and we can support their arbitration.
- sandworm101 10y ago
  If you plan on linking money to content, you better have a really definitive answer. That movement of money is essentially the purchase of a license. If a business or individual relies on your answer as to who that license should be purchased from, they do not want to be hit with a copyright complaint should the real owner appear. And if people end up abusing your system to monetize content that they should not, expect all sorts of legal docs in the mail.
  - parkan 10y ago
    Absolutely! Spotify's 30M settlement with NMPA is a great example of this kind of failure in existing systems