Ah, the potential is there alright. But until there has been attempts to solve it, concluding that it is not feasible or too hard to solve with current resources, it is just another solvable problem. That’s how I look at such things There’s no giving up on such things before. Do you think they would have tried to create “the impossible network” (SAFENetwork) otherwise?
There is a huge difference in the stance taken from the two sides here:
One says “let’s research how this can be designed to work, we don’t know all the problems before doing the work, but neither do we know of the solutions to potential problems before it.”
The other says “it’s impossible. Neither me nor anyone else has done the work of finding all problems or solutions, but I have this initial hunch, so everyone listen to me and stop working on this because I know.”
They are very different because one stance is open to whatever is actually found after researching it. The other is convinced it is impossible, without having actually put in the necessary work to be able to reach such conclusion.
Not really because if caching is working then even that attack is minimised and turns into something people might try but there is no real motivation since the resources used outweigh any gain. This was one of the most popular attacks discussed in this forum and the result is caching seems to be a reasonable effective solution so let test the shit out of it and see if we were right.
I’ve not speculated on this particular aspect before … IMO the individual upload reward may not be high, but the ROI looks really good on the face of it, as compared to farming - particularly if you have the money to market your uploads and drive traffic. So perhaps this is what people are considering as being a high reward - a high ROI (return on investment).
As said before I don’t mind that this stuff is going to be tried out, I’m not even involved in the actual work after all. It’s just that I personally can’t see a way around my stated arguments that would allow the GET rewards as described so far to work without glaring logical issues; i.e. I would like to know what I am missing if I am wrong.
Of course it’s fine too if no one is willing to waste further time trying to explain to me how my kind of argumentation is wrong, I could understand that. Anyway…
…I still don’t get this; if caching minimizes the attacks, does it not equally minimize genuine usage?
But that can’t be right, obviously it must not be allowed to minimize genuine usage to a degree to make genuine usage worthless.
But then that also can’t be right, because if genuine usage is rewarded by a noticeable amount, then automated abuse could easily be equally worthwhile (as stated in my previous post a botnet isn’t needed, or can someone perhaps counter that argument too?).
One way or the other there is a problem, if these statements are true.
Anyway, I will probably shut up about all of this now and take your word that it might possibly work…although I still would welcome further explanations & counter-arguments to mine of course!
Surely throwing pennies or less at something decent, at the click of a button or two, better rewards good content?
I am not a fan of engineering complex solutions, which may not even hit the target. I understand why people want this to work, but it seems open to abuse and of limited value when compared to the alternative.
Answered you before, and plenty of repeats in previous 500 odd posts (whew a lot)
The legit accesses come in randomly over location and time. Caching is not kept forever, but more frequent the more caching done. Slow attacks give virtually no returns since keeping a bot costs time/resources and/or money
Until you are willing to put some figures to the situation and see for yourself the lack of returns and thus incentive of these attacks you can keep thinking they will be quite significant.
You’re saying that with marketing you can get worthless material popular? Or that you can get material people find useful popular?
I mean, that first thing isn’t exactly easy to do. Sure possible, but it’s more like any remarkable success; practically everyone fail trying, very few succeed. If you succeed to successfully make people interested in objectively worthless material, over extended periods of time, with marketing that is compensated by the rewards - you have done something truly remarkable.
Because to abuse it you must attempt to get a LOT of GETs in order to justify the time, money and resources spent attempting to do it. No matter how, unless you get maybe a 100K people doing the GETS at random time. Anything too fast from say 10K or 20K people then caching will hurt you every time.
The problem is getting enough people in enough locations (XOR space wise) to make caching as effective as it it is for legit access. Good luck trying that and keeping your 100K friends and not raking up a lot of cost in personal time and extra expenses (pizza, phone bills, internet bills etc)
Doesn’t sound like sincere argument though, more like folding down, since you’re trying to say there shouldn’t be PtP. So I am guessing you don’t really have that super-idea how to get rich quick on it.
But then again, if you think you have. Sure
I’m not planning to get rich quick … just adding to residual.
I’m against PtP generally because of KISS principle and the fact that I don’t like the macro-economics of it … but if it is going to exist, then I will endeavor to take advantage of the micro-economics of it as much as is possible … perhaps it won’t be possible though. I think until it exists and we can get some solid estimates on it, then no way to know (for me personally).
My point is exactly that abusers could in fact hold the same properties as legit users, e.g. be spread in enough locations.
No, what I am thinking of is systematic, automated abuse, that can be easily joined by anyone.
E.g. a tool that sends random GET requests for resources of the automated abuser group (i.e. the data is simply discarded, the GET requests only serve to obtain the GET rewards for the abusers).
Thus there would be pointless work in the network that only helps the abusers. The more abusers join the abuse-system, the more the network as a whole will be affected. Yet individual abusers would benefit from the abuse (until there are too many abusers for the abuse to be sustained).
And such an abuse tool would require only the initial development and could have basically no further costs.
Very simplified. For one, it requires that you actually have any. Big selling point of network is that it’s free to use. Wouldn’t be that big of a thing if it wasn’t expected that many users will be using it without having balance.
So, a potentially genius influence on network dynamics is thrown away by neglecting this.
I don’t think anyone is. You research potential features to find simple solutions that hit the target. So…
They are not exclusive though so why put them up against eachother? About abuse, that is big part of design of such a thing naturally. For such potentially huge impacting things, “seems” is not quite enough. Just look at SAFENetwork, or any decentralised network, the quotation could very well have been said about it a couple of years ago (hell, some still do).
Imo the real potential of get rewards is to have an algorithm that dynamically adjusts farming, ptd, and ptp rates to maximize growth. It’s a multidimensional optimization issue. If there is not enough storage, the network raises farming ratio and reduce others… Network detects that not enough quality apps are being used, Ptd ratio goes up… Network detects that content consumption is down, Ptp rates go up and vice versa etc.
Obviously, coming up with ways to detect these things and writing the dynamic controls takes work and may turn out to offer little improvement over fixed ratios. So better to just start testing with fixed ratios like 1/10 as proposed. There is a lot of potential here.
I’m going to collect the few expressed concerns that I didn’t see already met here.
Owning audience will put you in a position to harvest quality content from skilled people without audience.
It might be true that they get the position, but does that equal that it is profitable?
How do you discover quality content produced by someone without audience? Seems like gold digger work, and that isn’t necessarily very profitable.
For it to be efficient it seems to me you can’t spend much time digging through piles of junk, to maybe find that undiscovered gem. You need to be made aware of it. The more likely that you become aware of it, the more likely the producer actually has an audience (the medium through which you got aware).
If however there are big audience whales trawling the mud to find undiscovered gems, I as an unknown super gifted producer, would actually be able to profit from being discovered and spread to this whale audience. Maybe that is a good trade off then. Because I don’t see it as a given that the audience would prefer the whale once they find out about this hard working gifted producer that they obviously like.
OK, let’s look at this one. It would be a fun tool to create!
So, I have junk content that I want to upload, but I need to generate traffic at highest possible rate, from enough varied locations, that I don’t hit an intermediary cache on the path.
There’s an upper limit to this rate. It looks to me as if the maximum number of nodes in last hop before destination, sets this limit. A bit simplified, you cannot exceed the number of requests it takes to round-robin through these before first called cache expires. To complicate it a bit, there’s the influence of churn, which introduce ‘artificial’ cache eviction in this set of nodes, at some frequency.
So for every chunk I have, I would want to find this optimal number of other users, that I exchange this chunk with for one of theirs, that I can periodically GET on, in exchange for them GETing on mine.
This tool can of course also be abused. It needs some credibility system, otherwise I could supply you with chunks but not care to do GETs on your chunks. I.e. cheating the tool. Considering the tool is a device for cheating a system, with target audience presumably being inclined cheaters, this is a quite high risk.
If it’s going to be a trusted group, the scalability and thus threat of the tool, is limited. If it is going to be a decentralised tool which anyone can join (as per the suggestion), then it needs to form these secure incorruptible groups of consensus that can make sure the rules are followed. Closest source of such logic would be SAFENetwork.
But already here, this seems like a very big project. Not an easy tool to develop.
Let’s ponder that for a minute before jumping on to further speculations.
Interesting, and hard to verify. But I guess there might be plenty of psychological research that come quite close - if not exactly spot on - to this. And that may very well be our best source of knowledge for it. Even after such a thing is in the wild, evaluation will be difficult due to the fact that a big part of the data to evaluate is protected by SAFENetwork.
Good counter point that a naive tool could itself be abused!
Hm so let’s instead say that the tool is somewhat more advanced and always “exchanges” GET requests between two parties, where each one fulfills the desired GET request of the other.
The parties randomly contact each other (Side note: The abuse system doesn’t have to be optimal, only good enough for most abusers long-term.).
To assure that the GET requests actually happen, e.g. some random part of the retrieved data could be checked. If the exchange is successful, a kind of “trust” score for these parties is increased.
This score could be stored both locally by the parties, as well as published globally for other tool users (cryptographically signed). To mitigate false published trust scores, the tool could only deem published scores themselves as trustworthy if the publisher is already directly or indirectly known to be trustworthy. </potential-somewhat-more-sophisticated-but-imperfect-tool>
tl;dr: You are right, a secure implementation would be non-trivial, albeit still feasible.
So I’m not completely convinced that this isn’t a potential serious issue long-term², but I no longer think it is extremely problematic, thanks!
As another side note, the content doesn’t necessarily have to be junk, only the abuser GET requests are “pointless”, but that doesn’t affect the arguments here anyway.