Thanks for the anti-ad for the ROM. Never heard of it, but now couldn't care less about it. Someone even helpfully suggests using torrents to help alleviate and instead gets roasted by the dev on...
Thanks for the anti-ad for the ROM. Never heard of it, but now couldn't care less about it. Someone even helpfully suggests using torrents to help alleviate and instead gets roasted by the dev on a rampage.
Guess what? Mitigating 90 TB of transfer a month is exactly what torrents help with. Hell, don't provide direct downloads at all. Post a torrent file with a bunch of public trackers, and let your userbase adapt. I download all my Linux ISOs using official torrent links if I can. It'll max out my gig internet in seconds, far faster than any direct download. Even a multiplexed one. It makes re-hosting the file trivial.
Complaining about direct download costs and "dickhead users" while simultaniously making it harder to reshare the files in question turns me off immediately. Whining about adblockers is passe and trivial to bypass: Self host your ads. You'll notice all the 'Download provided by' gets through my intense adblock filters just fine.
I'm reminded of some UI lesson about how your choices influence user behavior. If direct links are easy for users but mirrors and torrents are hard (especially if no shasums), then you only have yourself to blame.
If you have those kinds of storage and transfer needs you can probably co-locate a server with far more CPU, Memory, and storage than some VPS.
The whole post reeks of "I've tried nothing and I'm all out of ideas."
P2P file transfers and stuff are woefully underutilized in my opinion. It seems perfect for many open source and free projects like Wikipedia and OpenStreetMap... It's a bit of a shame. I mean, in...
P2P file transfers and stuff are woefully underutilized in my opinion. It seems perfect for many open source and free projects like Wikipedia and OpenStreetMap... It's a bit of a shame. I mean, in its current state torrents aren't really great for these things --- it's hard to update the swarm with new versions and stuff, but it seems like it wouldn't take much effort to improve that and make it all more viable. I guess that's what projects like IPFS are going for, but they don't really seem to be taking off that much either.
Of course Wikipedia and OSM both offer torrents, but they're snapshots of the content that get out of date and they're not really how the average person accesses those projects at all.
The problem is regular torrents do not generally scale down well IIRC. Below something like 50MB, the overhead of the tracker and peer exchange makes it less efficient than regular downloads. It...
The problem is regular torrents do not generally scale down well IIRC. Below something like 50MB, the overhead of the tracker and peer exchange makes it less efficient than regular downloads. It makes sense for snapshots and archives but not typical pageloads (video sites notwithstanding).
IPFS might be a solution like that. I do think the massive push to HTTPS has harmed efforts for caching content for resharing. I almost feel much of the push for HTTPS would have been better spent building up and simplifying PGP signing.
For vast quantities of content verification is more important than wire secrecy. Don't get me wrong, HTTPS was massively underutilized before, and overall mass adoption is a win.
Sadly being able to cache content for efficiency is almost directly at odds with preventing 3rd parties from snooping what you are doing. Though that snooping has largely been monetized by first parties these days so shrug?
I started going on a tangent there, but in short I'd love to see a more robust peer-exchange for regular web usage. Being able to donate bandwidth to your favorite sites would be an awesome way to sidestep the 'bandwidth costs money' problem a bit without requiring cash transfers.
Yeah, I agree, and I've been a little concerned about encryption leading to some inefficiencies with respect to caching and whatnot. I do wish IPFS and torrents had more privacy features baked in...
Yeah, I agree, and I've been a little concerned about encryption leading to some inefficiencies with respect to caching and whatnot. I do wish IPFS and torrents had more privacy features baked in --- it should be plausible to do something akin to onion routing, though I think it necessarily makes the network less efficient, and you'd need people to be willing to tunnel data for random downloads.
I'm not too familiar with the overheads for torrents. For large files like ROMs or map data, though, it seems like P2P is a clear win. You want to be able to provide a cheap mirror for many of these projects. It doesn't have to be fast, just low cost and available. The same is true for Linux distributions and their caches of packages. Obviously you want these files to be signed or something, but beyond that you mostly don't care where you get them from (maybe you want to leak what programs you depend upon to the fewest number of people possible, though, in which case you might care about privacy extensions to IPFS / torrents). I know there's been some talk of NixOS on IPFS, but I don't think anything has happened with that yet.
There's a couple of BEPs for it. I think one is just basically using RSS, and the other can update the contents of torrents via the DHT or something... I've never seen them used, but I could just...
There's a couple of BEPs for it. I think one is just basically using RSS, and the other can update the contents of torrents via the DHT or something... I've never seen them used, but I could just be uninformed.
As someone who actually wanted to implement functionality using self-updating torrents, it turned out that no major clients supported them. It doesn't help that they rely on another BEP that's...
As someone who actually wanted to implement functionality using self-updating torrents, it turned out that no major clients supported them. It doesn't help that they rely on another BEP that's vaguely defined.
Yeah, mid-blog he starts complaining about users for things they have no control and is really hostile to ideas suggested in comments. Also I disabled ad blocker on website to check and download...
whole post reeks of "I've tried nothing and I'm all out of ideas."
Yeah, mid-blog he starts complaining about users for things they have no control and is really hostile to ideas suggested in comments. Also I disabled ad blocker on website to check and download links are surrounded by ads.
Wow his entire reply about torrents sounds like nothing but excuses for not even attempting it. None of them make sense. ...Yes. Seeding something is not anywhere in the same realm as blocking...
Wow his entire reply about torrents sounds like nothing but excuses for not even attempting it. None of them make sense.
They know perfectly well that by using an adblocker we can’t even get the little add revenue we would otherwise might have gotten from the adds…
But they do it anyway…
Do you think those people are actually willing TO SEED and have their computer upload tens of GB’s only so they can help us?
...Yes. Seeding something is not anywhere in the same realm as blocking ads. I am militantly anti-advertisements but I am very open to seeding files. My computer is on all the time, my upload bandwidth does nothing to my internet use, I seed things all the time. In fact, I seed some stuff by accident because I forget about it. Pure leeching and ad blocking are not mutually inclusive in the slightest, I don't have any clue where this guy got this stupid idea.
It would actually cost them way more than allowing adds for ten minutes in their browsers and they know that as well…
How?
Have you ever tried to download a movie where there’s only one or two seeders who have set the upload rate to the most minimum and the rest are leechers?
I've been using public trackers for over a decade and never have run into this specific problem. If anything, it's not enough seeders and the only connection seeding is available overnight for me. Big whoop, my PC is on and downloading overnight.
I was trying to understand this and I think he might have been using torrents with now dead trackers, and a client that makes it hard to add extra trackers. When I moved to a different torrent...
I've been using public trackers for over a decade and never have run into this specific problem. If anything, it's not enough seeders and the only connection seeding is available overnight for me. Big whoop, my PC is on and downloading overnight.
I was trying to understand this and I think he might have been using torrents with now dead trackers, and a client that makes it hard to add extra trackers. When I moved to a different torrent client and had the ability to auto-add a few trackers I found that fixed a lot of problems I had been having.
I get this guy's frustration. Bandwidth isn't free, nor is it good when people use download managers and accelerators to bombard your server with requests. I do think he can reduce his hosting...
I get this guy's frustration. Bandwidth isn't free, nor is it good when people use download managers and accelerators to bombard your server with requests. I do think he can reduce his hosting costs considerably if he seriously thought hosting his stuff exclusively on Github (if they'll allow that), or switched exclusively to torrents. He also could have worded his argument a whole lot better rather than resorting to calling his own users assholes and dickheads through the power of Caps Lock.
Reminds me a little bit of a certain World of Warcraft addon site that a large portion of the WoW community have disavowed.
For those unfamiliar, Curse used to just host WoW addons but soon branched out into hosting all kinds of addons and mods across a variety of games, most popular being Minecraft. It was self-owned, then Twitch acquired it in 2016, and then they sold the company off to Overwolf about two years ago.
Even back in the earlier days of WoW, many used third-party addon managers to download addons rather than visit the official page, because they allowed one-click updates which were much quicker, easier and more conveient compared to visiting several webpages, downloading ZIP files, and extracting them into your addons folder. They also didn't use Curse's own official offering because they paywalled that functionality in their own official app.
Curse and Wowinterface previously raised controversy in 2009 by making moves to block Wowmatrix (one such third-party program) because it was taking up a lot of server bandwidth and ruining their ad-based model. But that drama died down quickly because both addon hosts kinda had a point when it came to how Wowmatrix deprived them of ad revenue. Also, Curse soon relaxed their policy and even gave third-party apps API access.
Curseforge's new owners, Overwolf, have made a lot of people in both the WoW and Minecraft modding communities disavow the website by blocking API access to third-party download managers and forcing people to install their shitty screen overlay software just to use their official client. Overwolf the program by the way is a memory hog and may-or-may-not be spyware, depending on how much you're willing to believe the hyperbolic and vitriolic reviews left for them on Trustpilot by infuriated WoW and Minecraft players. Fact that it is mandatory to install Overwolf just to use the Curseforge client, or the Feed The Beast (major MC mod) Launcher is a disgrace, regardless of how harmful the program really is to your PC.
Many people actually use Github, their own webpages, or Wago.io (a competing website) to host their addons now.
I personally don't trust Overwolf because it was once bundled with Teamspeak 3 without my knowledge nor consent (this was before the days where Discord was a widespread thing and when guilds still hosted their own TS voice chat servers.)
What an odd post... I couldn't resist fact checking it in the comments over there, so I'll be interested to see if I get a reply to this:
What an odd post... I couldn't resist fact checking it in the comments over there, so I'll be interested to see if I get a reply to this:
I sympathise with the frustration of dealing with demanding users, especially on a voluntary project, but the choice of examples here seems quite strange.
What would the advantage be of using a £50/month VPS to distribute 300GB of static files rather than just using object storage? The economies of scale there are pretty much unbeatable, especially given that you're worried about things like people opening multiple simultaneous connections: an S3 compatible store just abstracts that all away for you.
It looks like the site is currently hosted on Contabo, which charges just over £5/month for 500GB of object storage with (allegedly) unlimited data transfer. I can't comment on their specific fair use policy, but pricing is in line with Cloudflare R2, and I know that they won't blink at 90TB/month transfer.
If you did host 300GB on Cloudflare, it brings the two year cost example down from £1,200 (not £2,399 - that screenshot is for a four year plan) to under £90. Still generous to provide for free, no question about it, but an order of magnitude easier to cover.
If it were me I might consider a quick script to automatically generate torrent files when anything new is uploaded, with a web seed pointing back to the main storage URL, that way users have both options available and you can get some real-world data on whether it has an impact on bandwidth - but I can understand if the one off time cost of setting that up isn't a priority.
Yeah, I was trying to give the benefit of the doubt but his latest reply on another comment makes me think it's not in good faith: he says "some providers can be more expensive than other as you...
Yeah, I was trying to give the benefit of the doubt but his latest reply on another comment makes me think it's not in good faith: he says "some providers can be more expensive than other as you can see from the Hostinger’s example" and then further down implies that the average reader "who stops to look at our website even for a second" should realise they're actually with Contabo, which is absurd to expect. He knows that he specifically chose an example 5-10x more expensive than what they're actually using and somehow he's blaming the audience for that, while still claiming "the figures i presented in there are extremely conservative".
No idea whether my comment was one he considered "vitriolic", "snide", or a "good suggestion", but either way it didn't get posted. He fixed the two year/four year discrepancy in the post, at least.
I actually put that in as a direct counter to the fact that he was talking about some massive manual overhead to create them, and suggesting that lack of seeders was a dealbreaker, neither of...
I actually put that in as a direct counter to the fact that he was talking about some massive manual overhead to create them, and suggesting that lack of seeders was a dealbreaker, neither of which is true. I was trying to frame it as constructively and positively as possible, because unfortunately saying "you're talking out of your arse, mate" doesn't tend to do much to get people on side, but it doesn't seem to have helped.
Thanks for the anti-ad for the ROM. Never heard of it, but now couldn't care less about it. Someone even helpfully suggests using torrents to help alleviate and instead gets roasted by the dev on a rampage.
Guess what? Mitigating 90 TB of transfer a month is exactly what torrents help with. Hell, don't provide direct downloads at all. Post a torrent file with a bunch of public trackers, and let your userbase adapt. I download all my Linux ISOs using official torrent links if I can. It'll max out my gig internet in seconds, far faster than any direct download. Even a multiplexed one. It makes re-hosting the file trivial.
Complaining about direct download costs and "dickhead users" while simultaniously making it harder to reshare the files in question turns me off immediately. Whining about adblockers is passe and trivial to bypass: Self host your ads. You'll notice all the 'Download provided by' gets through my intense adblock filters just fine.
I'm reminded of some UI lesson about how your choices influence user behavior. If direct links are easy for users but mirrors and torrents are hard (especially if no shasums), then you only have yourself to blame.
If you have those kinds of storage and transfer needs you can probably co-locate a server with far more CPU, Memory, and storage than some VPS.
The whole post reeks of "I've tried nothing and I'm all out of ideas."
P2P file transfers and stuff are woefully underutilized in my opinion. It seems perfect for many open source and free projects like Wikipedia and OpenStreetMap... It's a bit of a shame. I mean, in its current state torrents aren't really great for these things --- it's hard to update the swarm with new versions and stuff, but it seems like it wouldn't take much effort to improve that and make it all more viable. I guess that's what projects like IPFS are going for, but they don't really seem to be taking off that much either.
Of course Wikipedia and OSM both offer torrents, but they're snapshots of the content that get out of date and they're not really how the average person accesses those projects at all.
The problem is regular torrents do not generally scale down well IIRC. Below something like 50MB, the overhead of the tracker and peer exchange makes it less efficient than regular downloads. It makes sense for snapshots and archives but not typical pageloads (video sites notwithstanding).
IPFS might be a solution like that. I do think the massive push to HTTPS has harmed efforts for caching content for resharing. I almost feel much of the push for HTTPS would have been better spent building up and simplifying PGP signing.
For vast quantities of content verification is more important than wire secrecy. Don't get me wrong, HTTPS was massively underutilized before, and overall mass adoption is a win.
Sadly being able to cache content for efficiency is almost directly at odds with preventing 3rd parties from snooping what you are doing. Though that snooping has largely been monetized by first parties these days so shrug?
I started going on a tangent there, but in short I'd love to see a more robust peer-exchange for regular web usage. Being able to donate bandwidth to your favorite sites would be an awesome way to sidestep the 'bandwidth costs money' problem a bit without requiring cash transfers.
Yeah, I agree, and I've been a little concerned about encryption leading to some inefficiencies with respect to caching and whatnot. I do wish IPFS and torrents had more privacy features baked in --- it should be plausible to do something akin to onion routing, though I think it necessarily makes the network less efficient, and you'd need people to be willing to tunnel data for random downloads.
I'm not too familiar with the overheads for torrents. For large files like ROMs or map data, though, it seems like P2P is a clear win. You want to be able to provide a cheap mirror for many of these projects. It doesn't have to be fast, just low cost and available. The same is true for Linux distributions and their caches of packages. Obviously you want these files to be signed or something, but beyond that you mostly don't care where you get them from (maybe you want to leak what programs you depend upon to the fewest number of people possible, though, in which case you might care about privacy extensions to IPFS / torrents). I know there's been some talk of NixOS on IPFS, but I don't think anything has happened with that yet.
There's a couple of BEPs for it. I think one is just basically using RSS, and the other can update the contents of torrents via the DHT or something... I've never seen them used, but I could just be uninformed.
As someone who actually wanted to implement functionality using self-updating torrents, it turned out that no major clients supported them. It doesn't help that they rely on another BEP that's vaguely defined.
Yeah, mid-blog he starts complaining about users for things they have no control and is really hostile to ideas suggested in comments. Also I disabled ad blocker on website to check and download links are surrounded by ads.
Wow his entire reply about torrents sounds like nothing but excuses for not even attempting it. None of them make sense.
...Yes. Seeding something is not anywhere in the same realm as blocking ads. I am militantly anti-advertisements but I am very open to seeding files. My computer is on all the time, my upload bandwidth does nothing to my internet use, I seed things all the time. In fact, I seed some stuff by accident because I forget about it. Pure leeching and ad blocking are not mutually inclusive in the slightest, I don't have any clue where this guy got this stupid idea.
How?
I've been using public trackers for over a decade and never have run into this specific problem. If anything, it's not enough seeders and the only connection seeding is available overnight for me. Big whoop, my PC is on and downloading overnight.
I was trying to understand this and I think he might have been using torrents with now dead trackers, and a client that makes it hard to add extra trackers. When I moved to a different torrent client and had the ability to auto-add a few trackers I found that fixed a lot of problems I had been having.
You have a point. I always download ISOs via torrent because it's faster.
I get this guy's frustration. Bandwidth isn't free, nor is it good when people use download managers and accelerators to bombard your server with requests. I do think he can reduce his hosting costs considerably if he seriously thought hosting his stuff exclusively on Github (if they'll allow that), or switched exclusively to torrents. He also could have worded his argument a whole lot better rather than resorting to calling his own users assholes and dickheads through the power of Caps Lock.
Reminds me a little bit of a certain World of Warcraft addon site that a large portion of the WoW community have disavowed.
For those unfamiliar, Curse used to just host WoW addons but soon branched out into hosting all kinds of addons and mods across a variety of games, most popular being Minecraft. It was self-owned, then Twitch acquired it in 2016, and then they sold the company off to Overwolf about two years ago.
Even back in the earlier days of WoW, many used third-party addon managers to download addons rather than visit the official page, because they allowed one-click updates which were much quicker, easier and more conveient compared to visiting several webpages, downloading ZIP files, and extracting them into your addons folder. They also didn't use Curse's own official offering because they paywalled that functionality in their own official app.
Curse and Wowinterface previously raised controversy in 2009 by making moves to block Wowmatrix (one such third-party program) because it was taking up a lot of server bandwidth and ruining their ad-based model. But that drama died down quickly because both addon hosts kinda had a point when it came to how Wowmatrix deprived them of ad revenue. Also, Curse soon relaxed their policy and even gave third-party apps API access.
Curseforge's new owners, Overwolf, have made a lot of people in both the WoW and Minecraft modding communities disavow the website by blocking API access to third-party download managers and forcing people to install their shitty screen overlay software just to use their official client. Overwolf the program by the way is a memory hog and may-or-may-not be spyware, depending on how much you're willing to believe the hyperbolic and vitriolic reviews left for them on Trustpilot by infuriated WoW and Minecraft players. Fact that it is mandatory to install Overwolf just to use the Curseforge client, or the Feed The Beast (major MC mod) Launcher is a disgrace, regardless of how harmful the program really is to your PC.
Many people actually use Github, their own webpages, or Wago.io (a competing website) to host their addons now.
I personally don't trust Overwolf because it was once bundled with Teamspeak 3 without my knowledge nor consent (this was before the days where Discord was a widespread thing and when guilds still hosted their own TS voice chat servers.)
What an odd post... I couldn't resist fact checking it in the comments over there, so I'll be interested to see if I get a reply to this:
Doubt it'll get approved, author seems quite combative. This seems more like he wants to rant than actually looking for solutions.
Yeah, I was trying to give the benefit of the doubt but his latest reply on another comment makes me think it's not in good faith: he says "some providers can be more expensive than other as you can see from the Hostinger’s example" and then further down implies that the average reader "who stops to look at our website even for a second" should realise they're actually with Contabo, which is absurd to expect. He knows that he specifically chose an example 5-10x more expensive than what they're actually using and somehow he's blaming the audience for that, while still claiming "the figures i presented in there are extremely conservative".
No idea whether my comment was one he considered "vitriolic", "snide", or a "good suggestion", but either way it didn't get posted. He fixed the two year/four year discrepancy in the post, at least.
Someone actually suggested using torrents and he basically started whining more.
I actually put that in as a direct counter to the fact that he was talking about some massive manual overhead to create them, and suggesting that lack of seeders was a dealbreaker, neither of which is true. I was trying to frame it as constructively and positively as possible, because unfortunately saying "you're talking out of your arse, mate" doesn't tend to do much to get people on side, but it doesn't seem to have helped.