Could watermarking data fix it all?

There’s a great debate going on here on the forum about paying the producer, or the App-maker, or both. In this topic I would like to purpose an idea that does both and makes implementing way more easy for the Devs I think. I already asked @dirvine a lot of questions about Farming and he gave a lot of answers and used the word watermark in his reply in this topic.

But here we go:

An RFC is introduced to “watermark” public data. The data could be anything as long as it’s public. This means a picture, a video, HTML, JAVA-script you name it. Whenever the data is called out of the network a Farmer might farm some Safecoin. So when I share a nice picture of my cat, and it goes to a great number of other websites on SAFE, I’ll earn 10% of the Farmed coins whenever the picture is requested and not in cache somewhere. Same for video, same for audio etc.

And what about the App-maker?? Well, the App comes to you as Chunks out of the network. And when the App is public data the App-maker could just watermark his App to earn 10% as well. So when thousands of people use an App to look at video, caching will kick in but the requests for the Chunks will come from all address range out of the network. This means that the requests come from maybe 32 places as seen from the Vaults’ point of view (after the requests went from HOP to HOP). The App-maker could make some good money on it! He could even change a little logo-picture on a daily base or so.

And let’s assume we have a video-App. You load the App, the Chunks create a GET so someone could farm and the App-maker makes some money. Next you load a video on that App and the uploader of that video made a watermark as well. So now he get’s paid because the App was used to load his video.


  • If you build a SAFE-website, and you use some nice pictures out of SAFE the creator of that picture get’s a little reward as someone is looking to the website you made.
  • You get some reward as well, because the HTML/CSS/JAVA is yours and was loaded out of the network and watermarked to you.
  • The network only allows people to watermark public data. So making thing public has an incentive for people to do.
  • If you look at a video-App or video-website on SAFE, people are rewarded the moment their stuff get’s requested.
  • Should be easyer to implement than all other systems with Apps that make their own money and prodecers have their own system etc.


  • One might create a 10 hour public video and put his own videoplayer at loop in the hope caching won’t kick in and make money on their own content. Although just normale Farming might make him just more.
  • Maybe there’s another way to trick this system, and we might never know when that’s done because we never know if people are requesting data just to make money on 10% of the farming.

But that’s it. Just make people able to watermark any data and we should be good. Both App-makers and content-makers

edit 1: Watermarking means > connecting a wallet-address to data. So people could even earn anonymous (think of a whistle blower who shows his company/government is corrupt etc.) One might be able to choose to attach a username or digital signature as well. But that should be optional, i.e. Satoshi Nakamoto could post 2 new whitepapers and sign them with his private key. At the same time he could attach a wallet-address.

edit 2: This is very close to the model that was purposed by David and others, but the difference here is that an App can’t make money by referencing/requesting other data. The way I get the old model is that an App could earn on “playing” a video that wasn’t owned by the App. So there was an incentive for the App to create a dataflow whatever it was. In this way the App could only make money on it’s own data when it’s loaded.


I thought that how it always was. Makes so much sense to me. (and I’m sure I’ve read it from various SAFE / @dirvine type sources in the past)

(I’ve been wondering what the furore was all about the last week, tbh)


@polpolrene Please can you explain what you mean by watermarking, and the differences/pros/cons relative to the PtP type of marking data with a wallet address.

I’m taking watermarking to be different to attaching a wallet address but I don’t understand exactly what you mean by watermarking - and would it differ for different kinds of data (how for an image, how for text etc?) And I don’t understand how this will differ from attaching a wallet address to data (though obviously it must).


1 Like

I support the intention… but…

I suspect it is possible to create multiple XOR connections (i.e. running many clients simultaneously). The exploiter could artificially simulate a mass demand of GETS, from different directions. This could be enhanced by disconnecting and reconnecting to get a different XOR address.

If this can be done, he/she would trick the Network into thinking there are many different people constantly GETTING the APP and CONTENT.

Please debunk this concern in a technical response.

Me too, but there’s a little difference. Read reply by David below:

That’s what David pointed out. But I’m not really sure if an App was needed in this example. My question was more like; There’s a SafeTube video App, I upload a video, it get’s a lot of plays. Will I earn money or the App??

Ohw yeah, sure. I made 2 edits on the post. Actually it’s the same. You could attach an address to all public data. Maybe with a real watermark of the uploader as well. Adele could post here new clip online, she makes money by people that watch the clip and maybe she should even have the option to attach a username with digital signature.

I whish I could :slight_smile: I asked David in the post mentioned above, this is his reply. This get’s so technical that I have nu clue what can be done and what not.

1 Like

@dirvine is referring to a caching threshold.

It’s like a circuit breaker, that trips when too much power (Repeat GETS) is demanded from the vault. Caching distributes the original vault burden by “temporarily” duplicating data chunks into other vaults. At the same time, it also cuts off all funding (Safecoin Rewards) for that particular chunk.

I think @dirvine is saying, “it’s more profitable to run multiple vaults compared to gaming watermarked content.”

This may actually be true if the bandwith cost for “repeatedly downloading your own App and Content” yields less Safecoin than using that same amount of bandwith to host multiple vaults. And the reason is the farming rate, which he already pointed out.


Being a long time programmer watermarking meant to me to be “modifying the data to mark it”. Rather like the watermarks on writing paper that was common in times past. Actually the US paper money has water marks on it (I think)

It took quite a while to shake that view when discussing this in relation to SAFE. The pay address is held in ?meta? data associated with the chunk and not in the data itself. It has to this way because the address has to be retrieved prior to decryption of the chunk, because the address is not returned with the data, but processed at the same time as the farming reward attempt.


hey Mr @polpolrene what about, just for now until all economics defaults get hammered out) if we JUST watermarked the public data, so that any donations will have a default wallet to go to?

could help facilitate the tipping economy …

Is watermarking going to happen 100% for sure yet? (as in assigning a wallet to all public data), even if PtP isn’t? THIS SHOULD AT LEAST HAPPEN, EVEN IF PTP ISNT 100% THERE YET

also do you have a link? i’d really love to read it but i can’t find it in any of the maidsafe github rfc folders

i think he means no app rewards (specifically), and SAFE only rewards farmers by default.

Other than that, ALL public data (apps, videos, txt files, html etc etc) is watermarked and it all earns 10%.

she he’s saying:
90% farmers
10& public data (watermarked)

and the old plan was:
90% farmers
10% app devs
0% everybody else

i think it’s an AWESOME solution to everything, considering the blurs between apps and websites and other things. this groups them all together and has incentives in all the correct logical places.

also watermarking would provide an amazing groundwork to for a tipping economy on top of this, straight out-of-the-box.


He is saying

100% farmers as happens with or without any other rewards
10% of the value of farming goes to watermarked (ptp address stored)
10% of the value of farming goes to apps (if the app stored/got it)

1 Like

watermarking was the method introduced to do PtP

The pay to address is stored with the data’s meta data (done as a public file is stored and if ptp pay address is given)


1, Does the watermark create traceability? It may be public data, but that doesn’t mean it isn’t illegal someplace.

2, Is 10% worth the trouble? We are talking about funding Music, Art, Science, Porn, Cinema, television, literature, scientific research etc. Is 10% of the network expense going to come even close to making a difference when split amongst all of those industries? I suspect it is a cool sales pitch but when you do the math it will be a massive waste of everybody’s time and money - Most could do better under the current system - It’s probably not worth the coding complexity. Especially if you try to spread that 10% amongst ALL of the content being produced.

3, Is it fair? Some media is going to cost a log more GETs than others. Is an hour long BDSM video worth 1000 times the SAFEcoin of a Satoshi Nakamoto white paper?

4, My main issue is that there is no counterbalance to the payment system. We assume that traffic is good, and we reward traffic. If traffic is really good it ought not need to be rewarded. If it is not good the network ought to have a way to make it expensive. I believe rewarding traffic is going to result in traffic, and traffic is a strain on even the best designed networks. Yes it is cached, but that means a massive portion of the machines on the network have downloaded the content. It may be fast, but it is not cheap.


No the watermark is striped at farming and processed for potential rewards. Does not progress far at all and is separate to the data path.

As an incentive then any amount is better than none.

No it is not going to fund research or such. But then the author can setup their web site to ask for more. But at least they will get something if others link their files separate to the author’s website/app

Thats up for discussions :slight_smile: Some say why is your desire not as worthy as another’s desire that you detest.

Lets test it @dirvine says that its going to take a few lines of code to implement. (testing will take a while anyhow)


I have been thinking safe network is too decentralized, my understanding is that any content data file is broken up into chunks and spread out onto the network in 4 copies.
Why can not have an option where one copy is store locally on safe network drive and have 3-4 copies spread out on the safe network.
Then the content creator could watermark there data and have a one in four chance collecting the farming rewards.
I think this would be possible my data could be transferred to straight a public folder on the safe network, where I could make changes to it and then have the file broadcast to the whole network.
So if a local community page had one copy on your peer, the network could cache the data locally in your area.

The idea that we can give a 10 percent rebate back to the unloaders via a popularity lottery system and people expect to make a living off of that is a joke.

Mostly it is a a false hype. I just think people are so excited about the idea, they don’t really think about the math. And then they are so attached to the idea that they want to engineer it in this way or that way, and fight over it tooth and nail, even though it really is pennies.

The network is very ignorant by design . It is a lousy judge of value - and there are way better ways to pay content than trying to engineer a 10 percent rebate.

Ummm I thought I implied that and said it will not fund things

Only if people keep saying its meant to make a living off for most uploaders

Lets test it and then the proof will be in the pudding, which ever way it happens.

For the few lines that @dirvine says it will take, its no skin off our noses to try it out in the testing phase

This can also be good, in that no one group discriminates against another.

And this has never been denied, and many including myself have said that having other methods does not preclude PtP and they can work together. At least the artist gets something when others link to the public file separate to any APPs tipping or otherwise.


Maybe if there was no cost involved. But these rewards will have to be paid for by 10% higher upload prices. That’s 10% extra incentive to not upload files as well. Content uploaders that don’t aim to please the masses are going to be worse off under this system, they’re not going to earn back that extra 10% they paid. PtP is a wealth redistribution scheme among uploaders, so there are going to be plenty of losers as well.



But wouldn’t it be wise to test it during test phase since @dirvine believes it to be just a few lines of code. Then we all can see evidence of expected costs and impact. Not perfect but should see some trends

Although saying 10% may be incorrect.

  • it could be a little too low if a massive amount of public data with watermarking is popular
  • it could be right on
  • it could be significantly too high if the ratio of GETs from public data with watermarking to GETs from other public data and private data is not that great.

Looking at successful farming GETs i.e. non cached
NOTE: this does not consider the fact that much of the public & private data will be rarely read, thus the PUT cost for that data would be an additional factor reducing the costs of PtP & PUTs in general

  • Let A == #GETs from public data with a PtP address watermark
  • Let B == #GETs from public data without watermark
  • Let C == #GETs from private data

And taking all other factors for PUT cost being the same when considering the ratios

Additional cost for PUTs is based on the ratio of PtP GETs to non PtP is

A / ( A + B + C )

Thus if PtP is 10% of the value of framing then the additional cost to support PtP has to be multiplied by the above ratio.

So if we have at the lower end where GETs for PtP data is equal to all other GETs then A == B + C then the additional %age is 5% extra and if its like A == 4 * (B+C) then the %age is 8% extra

Essentially it is never 10% extra unless all data GETs is PtP data.


Interesting idea!
All public data is watermarked by default. This removes one more speed bump for uploaders. This incentive encourages to me use the Network, even thought I have to pay to PUT.

One Condition
I think human feedback is necessary to assist the Network in determining rewards. The network can easily measure traffic, but doesn’t know if that traffic is (support or curiosity) by consumer.

One example could be a popular song that WILL get modified (to steal original uploader’s position) and re-uploaded to an aggregator where high traffic already exists… Basically a copy cat clone.

A (LIKE) Farming system could help distinguish the difference between the copy cat clone and the original. But if the people support the changes made (lyrics added), then they LIKE the copy cat clone and show their support.

Here’s how I see this working…
When a consumer “likes” a video or blog or whatever, they are sending a “+1 like” to the watermark address. Is this possible? If so, then each LIKE initiates a farming attempt for the watermark address, not the chunk. Regardless of the file size, the reward is determined by actual support.

Obviously, we must prevent/mitigate spamming from the same person. Otherwise the uploader would spam their watermark content for free farming attempts. I’m still working on possible solutions to prevent LIKE spamming. See below…

Q & A
What about multiple account creations?
It costs 1SC to “activate” an account, because LIKING is a PUT request. Keep in mind the LIKE is only a farm attempt, not a guaranteed payout.

What about disconnecting and reconnecting, getting a new XOR address?
If the LIKE is associated with my accounts ID, then XOR address doesn’t matter. This means I can only LIKE a watermark content once. Repeat likes would get deduplicated.

What about user collusion?
Yes, a group of users could agree to “like” each others content. They can even upload multiple 1MB content to generate more artificial LIKES (farming attempts). There are a few ways to mitigate it. See below…

Most ideal solution?
LIKING would cost 1 Safecoin. The SC gets recycled back into the Network, adding to its revenue. Spamming LIKES on your own content would result in a net loss, because you only get a farming attempt.

One drawback of this solution is when SC becomes more valuable (i.e. worth more $$$). Then 1 click liking becomes expensive. Divisibility would be required as well as a way to decrease how much SC it costs to submit a LIKE.

Why can’t a LIKE pay 1 SC directly to the content owner (watermark address)?
Because then the owner could LIKE their own content for free, making it look like it’s popular when it is not.