partial seeding improves your ratio - a myth

[Duckater]

The time sccale to seed to the same ratio would be different if was only torrent running as upload speed is slowed while downloading so the time scale if seeded at 100% time would take longer in proportion due to extra amount down loaded

I have 2 torrents running at the moment; a 9gb that is partially dl (75mb) and a 6gb that will be dl to completion. They are from different trackers of about the same speed and about the same numbers of peers.
The partially dl torrent is already finish but the other one is still downloading and will be completed in about 15 hr from now.

Let see if PB is right. If he is wrong I may get banned from this tracker for low ratio.

A lot of this will depend on seeder to leecher ratio as well as speeds of both sets

[jars]

I don't believe so, Duckater, I don't actually partial seed on ScT. I seed forever there and it takes an eternity for me to hit one one. While I do have a crap connection (fibre optic is arriving quite soon), ScT has free packs there's no need to partial seed there.

[renwickftw]

How can that be? If the partial seeder only uploads 5mb, then the other guy will have to upload 233mb. Would that be possible, such a big difference?

forget about the same buffer. i edited this. the ratio is the same, but the buffer for the whole downloader will be higher of course, because he downloaded more.

both offer a certain amount of chunks to the swarm. both have to download this certain amount before they can offer it. there will be a demand for these chunks in the 12 hours. one can offer 15mb, the other 700mb. since the conditions are exactly the same, both of them will end up with the same ratio. one has downloaded less, but can offer less. one has downloaded more, but can offer more.

it doesn't make a difference, that's why partial downloading doesn't improve your ratio compared to full downloading.

your eaxampe is very good by the way and hopfully makes people understand the logic.

Although I dont agree that they will be the same ratio, I understand where you are coming from. From my eyes, what you are saying is not the reality. Let's just hope that I'm wrong.

[Duckater]

Was joking jars m8
Only reason I let me sct account go a few yrs back was due to slow connection and was a choice between there and few other trackers.
Still could not answer wether I let the right account go but still have me ftn, fsc etc accounts which at the time I valued more and decided means could not use it properly would be better letting it go creating another place for some one else.
I am totally against trading so no way was it going to be traded

[jars]

Was joking jars m8
Only reason I let me sct account go a few yrs back was due to slow connection and was a choice between there and few other trackers.
Still could not answer wether I let the right account go but still have me ftn, fsc etc accounts which at the time I valued more and decided means could not use it properly would be better letting it go creating another place for some one else.
I am totally against trading so no way was it going to be traded

Cheers mate. My sarcastic meter is way off tap when torrent talk is involved.

[Polarbear]

I have 2 torrents running at the moment; a 9gb that is partially dl (75mb) and a 6gb that will be dl to completion. They are from different trackers of about the same speed and about the same numbers of peers.
The partially dl torrent is already finish but the other one is still downloading and will be completed in about 15 hr from now.

Let see if PB is right. If he is wrong I may get banned from this tracker for low ratio.

added: my max seed is 60/30kb

you can't compare two different torrents. even if they have the same number of peers, the swarm behaviour can be totally different depending on the speed of the peers. only if the conditions are the same you would see, that you will end up with the same ratio.


since the discussion couldn't convince me otherwise let me summarize my opinion:

regardless where or when you download a torrent. your file ratio won't be different when you partial seed in comparison with a full download. stop the torrent after a certain amount of time: when you fully download you'll have the whole content, a bigger buffer and the same ratio.

[TP635]

you can't compare two different torrents. even if they have the sam number of peers, the swarm behaviour can be totally different depending on the speed of the peers.

I understand that, We will see what happen in 24 hr.

[hitman51]

Polarbear...u are dumb or something...because they tell u to partial seed on ur home connections which suck! if u do that on a seedbox there no point obviously...but ur home connection like mine if its 40 kB/s up then partial seeding is sometimes necessary...understand?

[tknaught]

Your posts are normally pretty good Polarbear, but you are way off on this one. I can tell you from experience that partial seeding is a lot more effective than full seeding at building ratio. Your logic in the OP assumes that you don't have to compete with other seeders.

please explain to me how seeding only a portion of a file gives you an advantage in the swarm when you compete with other seeders.

Imagine that you are downloading a torrent that's divided into 6 blocks - call them Blocks A,B,C,D,E,F.

The partial downloader only downloads Block A. He is competing with one seeder who has all blocks. Assume that everyone's speeds are equal. If a new leecher comes on the swarm, he will see that there are two copies of Block A and one copy of Blocks B-F. He will request Block A from the guy who is partial seeding because that is the only block he has. In the end, the partial seeder has uploaded one block and has a ratio of 1 and the full seeder has uploaded 5 blocks and has a ratio of 0.83.

Now imagine the same scenario, however there is one partial downloader and 5 seeders. Each one of them will upload one block to the new leecher, however the partial seeder will have a ratio of 1, but the full seeders will have ratios of 0.17.

Now, in addition to this, because the partial seeder technically shows up as a downloader in the swarm, other leechers will be more likely to connect to him, since downloaders make both incoming and outgoing connections, whereas seeders only make incoming connections. This gives another advantage.

Now, this strategy can fall apart if a majority of the users try to be partial seeders, because they will probably all get the same blocks and have a lot of competition.

[mrnobody]

Partial seeding CAN (not necessarily 'will') be useful for new huge pack...with enough leechers.

the idea is that, for most connection, when downloading at max speed, the upload speed is diminished.


So, say if I download 1 gb out of 100 gb pack, from a connection like mine, it will take ~30 minutes. During that 30 mins, my connection will max out and my upload will be ~ 0 kb/s. However, once I am done downloading 1 gb, my download will go down to 0 kb/s (as i am done downloading) and i will be maxing out my upload instead, given that there is enough leecher. At max speed of my connection, it will take me like 2-3 hours to get my ratio to 1:1. In other words, it will take me 2-3 hrs to upload 1 gb back.

Now, during that time period, other leechers will likely be leeching at their max speed as well, hence their upload will be pretty low. Which kinda gives a partial seeder, who has partial portion of the torrent and will be maxing out his connection with it, an opportunity to upload.

BUT, partial seeding isn't good for long though. Once leechers have the portion of file you are partially seeding, your upload speed will fade out. For first hour or two, a partial seeder maxes out his connection (given that there is enough leechers) then his upload speed for that torrent will exponentially decrease.

edit: i just read it over, not sure whether what i wrote is understandable or not :-\