Can torrent trackers' credit (ratio) system be improved?

[SljWk3Js8]

All torrent trackers that I'm aware of currently base their "ratio", or their credit system, on throughput uploaded vs. downloaded.

The main problem with this system is that seeding is exponentially rewarded compared with uploading of new content. This disincentivizes users from uploading new material since they're basically treated like dogs. Users who put in very little effort to buy a seedbox or who happen to have a faster internet connection are treated like kings. Many torrent trackers give some sort of bonus for other activities, such as uploading new content, but that bonus always scales linearly and is usually quite small.

A secondary problem is that it this system is oblivious to the value of individual uploads. It treats a $2 CD on the same level as a $2 million rare item or a 10xLP Vinyl rip that took hours of effort to complete.

A tertiary problem is "freeleeches". This system allows most users to provide nothing in return for a long time, waiting for free handouts. Those free handouts are like the government printing money and giving it away -- they cause inflation in a sense, although it's not too relevant in market where all prices are fixed, and most credit is simply stored in users' accounts where it will never be spent (it's essentially lost).

As a consequence of their design, torrent trackers are generally a cesspool of the lowest order. Many users would rather copy the same dubious/low quality torrents from one tracker to another, re-encode other peoples' uploads, and so on, rather than spend a couple dollars to provide something of more value to the community. Most new material basically relies on charity from a small number of users, or on the misguided effort by newer users who haven't understood how the system screws them over, yet.

What torrent trackers have failed to do is appeal to higher class individuals. There are collectors with vast collections of material out there. There are many users who are willing to buy original content to share. But everyone is discouraged when they see hundreds of thousands of leeches living for free off of their work, and they're confronted with a system that provides no incentive for them to work. It's analogous to how few people would want to work if they could live off of government welfare.

My question asks about "improvement", and this is obviously subjective. Currently, trackers are very much quantity over quality and are designed to welcome and appeal to the lowest members. What I'm wondering is if a different credit system could be invented that would reverse this.

The first consideration is how to tackle the main problem above. Here are things that I can think of that ought to be incentivized in a healthy sharing community:
1. Uploading new content
2. Seeding existing content (throughput)
3. Keeping torrents alive (seeding # of torrents; re-seeding)
Current systems treat #2 as most important and #1 and #3 as small bonuses. I am thinking about a system that treats #1 as most important and #2 and #3 as of lesser (or equal) importance. It's possible that #2 and #3 could simply be expected things to do based on morality, or failing to them could be punished.

The second consideration is for the value of an upload. I suggest that it needs to be quantified based on factors such as its market price, rarity, and the difficulty/amount of energy that went into sharing it.

The third consideration is how to design the credit system. I haven't figured this out yet, because it's a tricky problem. Current torrent trackers inject new credit into the system via new user accounts and freeleeches. Without credit in the system, things would grind to a halt. Some credit leaves the system as accounts are closed or abandoned, and credit becomes centralized in members that give more to the community than they take. The amount of credit floating around determines how much activity can be going on in the system at once.

Here are some thought experiments:

A) The principle of unlimited sharing. Say you have 100 users and each share 1 upload. Each user would get 99 downloads.

B) Add value. If 1 user contributes 5 units of value and 99 contribute 10 units, the 1 user would get 495 units.

C) Sharing is not unlimited. If only 10 users are interested in a user's upload, the user should only get 50 units of credit (5 from each of them).

D) The system starts with 0 credit. It has to be created before the first transaction can take place.

E) In a few decades, torrent trackers may have central banks, because the credit system is basically a currency. Perhaps re-seeding a dead torrent will be at a cost negotiated between the two parties. :-)

My final consideration is the moral one. BitTorrent is a technology that could be used in many different economies. It's wrong to associate it with one particular economy. Yet in a world where many people believe that universal, indiscriminate sharing is the "moral" thing to do, it may be very hard to operate a community in defiance of that principle. Prospective members would need to be filtered. A certain level of trust is required. The fact that digital information can be replicated indefinitely works to undermine that trust.

Any economists in here? Anyone interested in creating a better torrent site, if we can bash out the details?

[zora]

Interesting ideas but I see one major flaw.

There IS an incentive to upload content more rare then others. It's called the request system. People can request stuff to be uploaded and offer a bounty for it to be done. Now lets say there's a rare movie that is not uploaded yet. Someone may request it and offer a 10gb bounty for it. IF it's really sought after, more people can add to the bounty. Depending on the rarity, you could end up getting a reward of a couple hundred gb for the upload. However, if on the other hand something is fairly easy to come by (a new movie release etc.), the reward given for the uploaded content is a lot lower. This system rewards uploaders on giving the community what they want.

Another problem is this. If we put uploading content as the biggest priority, we will be putting quantity of uploads over download speeds. I'd personally like to have a tracker with less torrents but great speeds over one with lots torrents but only 2-3 seeders.

Also who/what determines rarity? Price of item? How many are around the world? Action Comics #1 is rare and expensive but it's pretty easy to find a copy of it to read online. It's an arbitrary term that would have to monitored by an individual person/group and frankly I don't think that's a good use of anyone's time.

You might of addressed some of these points but I just quickly skimmed your post during one of my study breaks. I might come back to it later tonight and dissect your post a little more.

[Instab]

There IS an incentive to upload content more rare then others. It's called the request system. People can request stuff to be uploaded and offer a bounty for it to be done.

that requires requests to be made first. it doesn't help uploaders who wanna upload quality content on their own.

I'd personally like to have a tracker with less torrents but great speeds over one with lots torrents but only 2-3 seeders.

what are great speeds good for if it's not what you want?
and what if those 3 seeders all have gbit servers? still too slow then?

[SljWk3Js8]

Interesting ideas but I see one major flaw.

There IS an incentive to upload content more rare then others. It's called the request system. People can request stuff to be uploaded and offer a bounty for it to be done. Now lets say there's a rare movie that is not uploaded yet. Someone may request it and offer a 10gb bounty for it. IF it's really sought after, more people can add to the bounty. Depending on the rarity, you could end up getting a reward of a couple hundred gb for the upload. However, if on the other hand something is fairly easy to come by (a new movie release etc.), the reward given for the uploaded content is a lot lower. This system rewards uploaders on giving the community what they want.

A few quick thoughts (I have to run as well)...

- The request system ties up credit indefinitely. The amount of request bounty that can be in place, and hence the number of requests in place, is limited. It would be better if bounty was a pledge that was calculated dynamically, so users could pledge more than their total credit without it being tied up unfilled.

- Requests have to be created manually which many don't bother with. It's not automated at all. Talk about value being "arbitrary" and requiring a lot of effort. I'm talking about using a heuristic to automatically assign value to every release (which could be manually adjusted when necessary).

- Request bounty is only a very crude estimate of the value of a release to the community. For instance, the # of users who snatch it would be a much more accurate estimate!

- It's still the kings with the seedboxes who are dictating requests rather than the kings who provide the most content. Also, if the upload is popular enough, those kings with the seedboxes will make back much of their request bounty by re-seeding the upload to hundreds of others. As I pointed out, the system is greatly stacked in their favor.

Also who/what determines rarity? Price of item? How many are around the world? Action Comics #1 is rare and expensive but it's pretty easy to find a copy of it to read online. It's an arbitrary term that would have to monitored by an individual person/group and frankly I don't think that's a good use of anyone's time.

If you know that a release costs $X and Y amount of effort to procure, then you have a simple formula for its value. Any user could buy it and spend the time to rip it themselves for that cost, or they could snatch it from the site for that cost. Corner cases would be harder and require some manual intervention.

[zora]

that requires requests to be made first. it doesn't help uploaders who wanna upload quality content on their own.

Yea but if you reward people for uploading instead of seeding. I think you will see a flood of garbage quality uploads just for the sake of getting these points so you can download what you really want.

what are great speeds good for if it's not what you want?
and what if those 3 seeders all have gbit servers? still too slow then?

You are assuming people will still see a need for seedboxes when they aren't rewarded as much for seeding then just uploading.

A few quick thoughts (I have to run as well)...

- The request system ties up credit indefinitely. The amount of request bounty that can be in place, and hence the number of requests in place, is limited. It would be better if bounty was a pledge that was calculated dynamically, so users could pledge more than their total credit without it being tied up unfilled.

Unless I'm misunderstanding you what I'm getting from this is that you would like to give people the ability to pledge say 100 points when they only have 75? How does that make any sense? How are you certain the person will ever pay the "dept" back? If a request is fulfilled does the individual now have -25 points?

It's true that the amount that you can offer up for bounty is tied up to your current ratio/upload credit. But you need to give something up to gain something. The individual making the request has uploaded X amount and he needs to decide how much of that credit he has accumulated to others are they willing to pay for their specific request.

- Requests have to be created manually which many don't bother with. It's not automated at all. Talk about value being "arbitrary" and requiring a lot of effort. I'm talking about using a heuristic to automatically assign value to every release (which could be manually adjusted when necessary).

Making a request is not time consuming. It takes 5 minutes at most. If one is too lazy to take 5 minutes to make a request, I'd argue they do not want the item enough. There is no possible way to automatically assign a value to EVERY release possible. Sure a system could probably catch all the major weekly music releases but what about that random indie band from some town in the middle of no where that no one has heard about? Their music record is probably more rare but chances are no one would care about the upload anyways. A value would have to be assigned manually and who can say how much it's worth? There's no way for someone to know everything about every musician ever so an objective grade would never be possible. Maybe the band turns out to be an underground hit, or it flops. But if someone really wanted this album they could request it.

True, maybe someone uploads the cd before a request is made and gets no bonus for it. But they choose to upload it. I really don't see why there always has to be some sort of incentive for people to do something. Sometimes people just want to give back to the community?

- Request bounty is only a very crude estimate of the value of a release to the community. For instance, the # of users who snatch it would be a much more accurate estimate!

I agree. But lets assume you are uploading on a really bad dial up connection. Even if your internet is absolute garbage, you are the person who originally uploaded the file so you are pretty much guaranteed to get as much upload credit as large/small as the file size. Sure at this point you will probably start being beat out by seedboxes and your contribution is useless. But if this really the issue, just implement a system lik cbt or 32p where your ratio is meaningless but you get points per hour for seedtime (the longer you seed something the higher you accumulate your points). The only problem with these systems is that it's so easy to cheat on them. All you gotta do is download like 30 torrent files, open mR and set all the files to seeding. After a week, you will have so many points you won't know what to do with all of them and you did zip to helping the community.

- It's still the kings with the seedboxes who are dictating requests rather than the kings who provide the most content. Also, if the upload is popular enough, those kings with the seedboxes will make back much of their request bounty by re-seeding the upload to hundreds of others. As I pointed out, the system is greatly stacked in their favor.

I know of no heavy uploader on any tracker who is having a hard time maintaining a ratio. True this is very anecdotal and those arguments are weak but truth be told there are only three real ways of surviving on a tracker.

1) Cheat (not really helping the community but at least it's free)
2) Use google fibre (or an equivalent fast speed internet) (super expensive monthly rate)
3) Buy a seedbox (probably the most moral and cheap option of the three)

Any of those 3 options should give you enough upload credit to throw away upload credit on requests if you really want something. If you're trying to survive on a private internet tracker without doing any of those above options, your doing it wrong.

If you know that a release costs $X and Y amount of effort to procure, then you have a simple formula for its value. Any user could buy it and spend the time to rip it themselves for that cost, or they could snatch it from the site for that cost. Corner cases would be harder and require some manual intervention.

Manual intervention breaks the system. As I pointed out, there's no way something can know how much everything is worth and what it took to get it. And your formula isn't as perfect as you may think. Lets take a random movie and pretend that it's available on blu ray and on iTunes. The blu ray version you need to go out buy it, rip it and upload it. The iTunes one you just stay at home and rip the DRM and upload it. The level of effort is probably less for getting an iTunes version of the movie. But I personally prefer iTunes rips just because they are usually smaller in size. Your formula doesn't account for preference. But with that being said, you can't account for preference because of seedboxes. Seedboxes just download everything that is uploaded to a tracker so a file could get let's say 5,000 downloads but a small fraction of those are people who will actually enjoy the content.

[Instab]

Yea but if you reward people for uploading instead of seeding. I think you will see a flood of garbage quality uploads just for the sake of getting these points so you can download what you really want.

no, the reward will be based on the "value" of the upload as you suggested

You are assuming people will still see a need for seedboxes when they aren't rewarded as much for seeding then just uploading

i am and they will. seedboxes can be used for many things. but either way a tracker without seedboxes doesn't last because the majority of all home connections worldwide is just too poor

[SljWk3Js8]

Yea but if you reward people for uploading instead of seeding. I think you will see a flood of garbage quality uploads just for the sake of getting these points so you can download what you really want.

Not necessarily. Uploads could be made worth nothing at all if nobody snatches them, for example. This thread is for thinking outside the box about all the possibilities. Please try to do that before simply proclaiming that it would not work.

Personally, I would prefer to only allow high quality uploads of new content on my torrent site. Scavengers copying junk from elsewhere on the internet are just wasting time, in my opinion. That kind of behavior is popular on traditional torrent sites, but only because they are wastelands with low quality standards.

Unless I'm misunderstanding you what I'm getting from this is that you would like to give people the ability to pledge say 100 points when they only have 75? How does that make any sense? How are you certain the person will ever pay the "dept" back? If a request is fulfilled does the individual now have -25 points?

I was thinking that you could pledge 100 points for a request and if you only have 75 at the moment, then your effective pledge would only be 75 were it to be filled right now. The key point is that you could pledge 75 points on multiple different requests rather than tying it all up in a single request that may never be filled (like a frozen bank account). Debt is an option I hadn't thought of, and it would give a whole new meaning to the "one account" rule at what.cd. If one request is filled and you go down to 0 points, your effective pledge on all the others would go to zero until you earned more credit somehow.

I just mentioned this as a possible way to overcome one of the major limitations with existing requests systems, so let's not get too far off topic with it.

Making a request is not time consuming. It takes 5 minutes at most. If one is too lazy to take 5 minutes to make a request, I'd argue they do not want the item enough.

So if you want 1000 items, you'd have to spend 83 hours (the better part of a week) making requests. Yea, that's a major waste of time, which is why most items tend not to have requests for them even if someone would be interested in having the item were it to be uploaded. That's not to mention that you might not know about many releases in advance, but still be interested in them once you find them uploaded. There are many people who maintain large collections of downloaded music, so I think it's safe to assume that most items would be desirable to at least a couple users.

I really don't see the point in arguing that every release should have a request made for it so that the filler can get the market price via a bounty. That's just adding a whole level of manual work (creating and filling requests) on top of a system which only roughly approximate the value of a release to a community. Requests will only realistically be made for a fraction of all possible music items. My heuristic for automatically determining value would be more accurate and cover a much larger amount of items and be automated and consistent.

There is no possible way to automatically assign a value to EVERY release possible. Sure a system could probably catch all the major weekly music releases but what about that random indie band from some town in the middle of no where that no one has heard about? Their music record is probably more rare but chances are no one would care about the upload anyways. A value would have to be assigned manually and who can say how much it's worth? There's no way for someone to know everything about every musician ever so an objective grade would never be possible.

Don't be such a naysayer. If you can automatically query Amazon, eBay, Discogs, bandcamp, and junodownload, you can get an estimate of the price for the vast majority of music. If you want to exclude outliers that can't be found, you can simply disallow uploading of music that's not found in one of those databases. Easy.

True, maybe someone uploads the cd before a request is made and gets no bonus for it. But they choose to upload it. I really don't see why there always has to be some sort of incentive for people to do something. Sometimes people just want to give back to the community?

The point is that most people don't want to give back to the community, which is why what.cd, for example, only has a tiny fraction of available music on it, much of it in poor quality uploads. Of their 100K+ members, only a tiny fraction are demonstrably interested in giving back out of the goodness of their hearts. So you personally may love torrent systems as they are and may love giving back to the community, but you are an anomaly if that is the case. This thread isn't about you in particular, but about how to design a system that incentivizes more, less-charitable users to contribute.

I agree. But lets assume you are uploading on a really bad dial up connection. Even if your internet is absolute garbage, you are the person who originally uploaded the file so you are pretty much guaranteed to get as much upload credit as large/small as the file size.

I didn't say that the original uploader gets nothing for their upload. I said that fast seeders are rewarded exponentially more than the original uploader, which is true. If you upload something and there are 10 leechers, you can expect to get ~1X the size in credit and people with seedboxes will get ~9X the size in credit. If there are 100 leechers, you'll still get ~1X the size in credit and they'll get ~99X the size in credit. They get a feast from your dollars and labor, and you get thrown a bone. They're like the CEOs and you're getting paid hourly wage. Why should you continue to make them rich when you're the one doing most of the work? Duh!

Sure at this point you will probably start being beat out by seedboxes and your contribution is useless. But if this really the issue, just implement a system lik cbt or 32p where your ratio is meaningless but you get points per hour for seedtime (the longer you seed something the higher you accumulate your points). The only problem with these systems is that it's so easy to cheat on them. All you gotta do is download like 30 torrent files, open mR and set all the files to seeding. After a week, you will have so many points you won't know what to do with all of them and you did zip to helping the community.

In bold: you refuted your own argument there.

As I said in the OP, there are three things I would like to reward in a healthy torrent sharing community. You're again proposing a system that only rewards one of the three (and then go on to explain the obvious flaw in the system you proposed).

I know of no heavy uploader on any tracker who is having a hard time maintaining a ratio.

Sure, if you upload enough items you'll be fine. But that doesn't make it fair or a system that most people would be interested in contributing to. If others pulled their weight, you should only have to upload a fraction of that amount to survive.

If you can either upload 1000 releases (say the cost is $1000 + 1000 hours of your time) or pay $100 for a seedbox and the end result is the same, then most users would prefer to spend the $100. That's called disincentivizing uploading new content.

True this is very anecdotal and those arguments are weak but truth be told there are only three real ways of surviving on a tracker.

1) Cheat (not really helping the community but at least it's free)
2) Use google fibre (or an equivalent fast speed internet) (super expensive monthly rate)
3) Buy a seedbox (probably the most moral and cheap option of the three)

Any of those 3 options should give you enough upload credit to throw away upload credit on requests if you really want something. If you're trying to survive on a private internet tracker without doing any of those above options, your doing it wrong.

On a traditional torrent tracker, yes, and I'm well aware of that. But torrent trackers don't inherently need to be that way. This thread is to consider and discuss possible alternatives. You seem only to be interested in defending the existing system, though....

Manual intervention breaks the system. As I pointed out, there's no way something can know how much everything is worth and what it took to get it. And your formula isn't as perfect as you may think. Lets take a random movie and pretend that it's available on blu ray and on iTunes. The blu ray version you need to go out buy it, rip it and upload it. The iTunes one you just stay at home and rip the DRM and upload it. The level of effort is probably less for getting an iTunes version of the movie. But I personally prefer iTunes rips just because they are usually smaller in size. Your formula doesn't account for preference.

I don't understand the problem. If the user has to identify the release as coming from a BD or iTunes when they upload it, why couldn't we value it differently accordingly? If the iTunes version is smaller, it's probably a different encoding and would be uploaded as a separate release from the BD version, if both encodings are allowed on the site.

My idea of quantifying the value for each item is to quantify the value of that item to the community, not necessarily the value that the individual paid for it if they made a mistake and overpaid. If there are 2 ways to get the exact same item and one is significantly cheaper than the other, then it wouldn't be very smart to use the more expensive option, would it?

But with that being said, you can't account for preference because of seedboxes. Seedboxes just download everything that is uploaded to a tracker so a file could get let's say 5,000 downloads but a small fraction of those are people who will actually enjoy the content.

The seedbox strategy of downloading things indiscriminately only works on traditional torrent sites where seedboxes earn credit by seeding and use that credit to download more items, use them to earn even more credit, and so on. That strategy would not work on a tracker that required uploading new content (something a seedbox itself can't do) to earn credit. So this is a null argument.

[anon]

Not very related to the discussion at hand, but I wanted to point this out:

(...) lets assume you are uploading on a really bad dial up connection. Even if your internet is absolute garbage, you are the person who originally uploaded the file so you are pretty much guaranteed to get as much upload credit as large/small as the file size.

That's true. And those who don't cheat nor have the leverage to fix the world will settle for it. Upload a single copy, let the seedboxers fuck you over, then possibly keep on seeding nonetheless in hopes of getting more down the road, once the others have deleted your torrent from their box's disk.

A common argument is that certain trackers "have everything" and thus it's impossible to find something to upload, but that is incorrect according my experience. What is true, however, is that doing all this so that you can get stuff you want at some indeterminate point in the future isn't so appealing when your free time is limited.

Also, "guaranteed" is a stretchable word. When I uploaded on What.cd, no one grabbed most of my music.

[SljWk3Js8]

Here's a simple example of a possible, working credit system:

Uploader's Crux

1. Every time a user uploads a torrent, they receive 1X its value in credit.
2. Every time a user downloads a torrent, they transfer 1X its cost in credit to the person who uploaded the torrent.
3. Assume that users are required by rules to keep all of their torrents (and downloads) seeded.

Pros:

- Credit is injected into the system in a way that promotes users to do something valuable, rather than giving it away for free (which promotes freeloading).
- Uploading a torrent is highly incentivized. Indeed, it's the only way to survive or to do anything in this system.

Criticisms:

- Users who happened to be the first to upload popular torrents are rewarded indefinitely when someone downloads it even years down the road. Is this fair?
- If a single user pooled too much credit, the system could stagnate. Especially if they uploaded thousands of items.

Possible improvements:

- Maybe "ownership" of a torrent (who receives the credit when someone purchases it) should be phased away from the original uploader over time. That credit could go to the community (ie. to the most active members or something), or it could go to the seeders who keep it alive, or such.

---------- Post Merged at 17:46 ---------- Previous Post was at 17:19 ----------

Here is the above combined with eliminating credit. Think of the credit-free system as being based on direct trading, with no currency involved.

Credit-Free Uploader's Crux

1. Users indicate their interest in downloading specific items with a "request".
2. The system searches for cycles in the graph of users' requests. Reducing a cycle will result in at least 1 user getting to download one of their requests.
3. Users will have a "credit" score behind the scenes to deal with the fact that not all releases cost the same amount of credit. Otherwise, cycles could only be reduced if they were zero-sum, which isn't very realistic. As a corollary, a request for an item could only be filled if the uploader of that item had requests, and if the requester of that item had an upload that someone else requested.

Pros:

- This system would eliminate the need to worry about injecting credit and managing about how much credit is in the system. It's basically akin to using direct swapping of goods instead of using a currency. The problem a currency is deciding how much to create and how to add it to the system.

Cons:

- There would be no immediate downloading. It could take some time to get what you wanted (unless your request happened to complete a cycle).
- If a user has no requests, it would be impossible for anyone else to get at their uploads.
- It may be too fair.
- Combining it with the Uploader's Crux, it would exacerbate the problem of the first uploader of a popular item.

---------- Post Merged at 18:02 ---------- Previous Post was at 17:19 ----------

Here's a modified version of Uploader's Crux.

Uploader's Advantage

1. Every time a user uploads a torrent, they receive 1X its value in credit.
2. Every time a user downloads a torrent, they transfer 1X its cost in credit which is divided among the the users who "own" the torrent.
3. The users who "own" the torrent is no longer solely the uploader. It now starts as solely the uploader, and then is gradually skewed over time towards the seeders. If the uploader remains an active seeder, they will still get the largest chunk.

Pros:

- Credit is injected into the system in a way that promotes users to do something valuable, rather than giving it away for free (which promotes freeloading).
- Uploading a torrent is highly incentivized. Indeed, it's the only way to survive or to do anything in this system.
- Seeding torrents is incentivized, because it ownership gradually transitions from the uploader to the seeders (which will usually include the uploader).

Criticisms:

- If a single user pooled too much credit, the system could stagnate. This danger is now spread across both major uploaders and major long-time seeders.


Other notes:

- This tweak would have interesting implications for the Credit-Free version as now a match doesn't have to be made solely between uploaders, but can go through seeders as well, which would make finding matches much easier.