On 28/08/2018 13:55, Michael Stone wrote: > On Tue, Aug 28, 2018 at 01:16:45PM +0100, Mark Rousell wrote: >> NNTP was inefficient in this regard compared to what other protocol or >> protocols, exactly? > > FTP and later HTTP, which handled binaries efficiently. In fact, one > was even named in a way to suggest it was a good way to transfer > files. :)
HTTP came later and wasn't relevant in the timeframe to which you referred. Additionally, both FTP and HTTP are not federated, many-to-many services or systems. I say again that Usenet was unique in this timeframe for the use case of public access, one-to-many, binary distribution. Except for perhaps hacked servers in some cases, FTP never did have much of a part to play in binaries distribution from what I could see. I think it was file sharing P2P protocols that eventually reduced people's preference for Usenet coupled with (as you say) ISPs' great difficulty in continuing to support Usenet servers. > Yes, academic, commercial ISP, and paid subscription servers. I also had some insight into what it took > to keep the servers running, not just the user-side view... I followed a number of text groups, until the > signal to noise ratio got low enough to make it not worth the effort. I too was a Usenet user in that timeframe and worked for an ISP at the time, although not directly on Usenet/NNTP servers. > You seem to have an overly idealistic view of the level of logging on > most news servers 20 years ago. You mean about the same amount of logging as on mail servers, FTP servers, or anything else at the time, then? Sure, there wasn't much logging in practice. I didn't say there was. I'm not being idealistic. I am simply observing that, logging or not, accessing a NNTP server did not hide one's IP address any more then than it does now. Indeed, despite greater legislation-mandated logging in many countries, the technical opportunities to access a server of potentially any type in a genuinely anonymous way are much greater now than they were back then due to widespread availability of VPN services! I therefore do not agree that anonymity was a primary driving factor for the use of Usenet for one-to-many distribution of binaries (although I don't doubt that the essentially false idea of anonymity may have influenced many less-expert users). I'm not being idealistic about the amount of access logging that went on when I say this; I am simply being pragmatic. I am being pragmatic because Usenet was simply the only widely available, worldwide, federated, public system available to distribute data (especially binaries) in a one-to-many manner. Other systems or protocols such as FTP just couldn't do what Usenet could do back then. > Also, for the record, I don't think I ever had a "training course" on > usenet. That's good. :-) > As far as being wrong...if LE siezed an anonymous FTP server > distributing illegal content and either reviewed its logs or monitored > its link they could get a list of each IP that accessed content. And the very same applied to a NNTP server attached to Usenet (or a standalone NNTP server for that matter). It was and is no more difficult for a NNTP server than for a FTP server, or an email server, or anything else. > There is no central point from which you can see who accessed usenet > content. But why would you expect there to be? It's a federated system. If you were expecting such a thing then you were expecting the wrong thing. One might also observe that looking for who accessed Usenet content is surely a waste of time. If one is interested in preventing distribution of illegal data of some sort then the primary concern is the sender, and the sender was not anonymous with NNTP (regardless of the existence of logs or not). Remember, the point here is one-to-many distribution, and it's the one that law enforcement should surely be interested in. > The bottom line is that for a period of time, usenet was the easiest > way to obtain certain illegal content. There were certainly overblown > reports that usenet was nothing but illegal content, and it's > certainly possible to transfer illegal content via other protocols, > but it's naive and/or disingenous to pretend that usenet didn't have a > problem. Oh I agree with you on this about Usenet. But: (a) I don't blame NNTP for this since it is not responsible for Usenet's problems that ultimately derived from its massive scale, not its protocols. The problems occurred as a result of the fact that Usenet was and is a massive, worldwide, publicly federated one-to-many distribution system. I.e. It was custom made (without its creators even realising it!) for large scale distribution of binaries, whether legal or illegal. (b) I dispute that Usenet had any real, reliable anonymity (although I accept that some people may have erroneously believed it did). I have no idealism or inflated idea about the level of logging that was common at the time on any server, regardless of the protocol it served. I take the view I do simply because every access to Usenet had to take place via a server which was and is, both then and now, the natural choke point for law enforcement to investigate senders of data deemed illegal over any protocol[1]. Furthermore, the false belief of anonymity paled into insignificance (in terms of Usenet's success as a binaries transfer medium) in comparison to the fundamental and unique one-to-many nature of Usenet, in my view, experience and memory. Yes, the bandwidth needs of Usenet was a problem for ISPs to cope with but that wasn't users' direct problem in terms of choosing to use Usenet for one-to-many binaries distribution. > The one to many capability simply didn't outweigh the enormous volume > of one-to-none-via-many. Even back in the day there were a lot of > really passionate advocates of the theoretical greatness of the > service, with no clue of how much it was costing to provide. I agree. This is a problem with Usenet that I have not disputed. It's just that, coming back to the original subject of this discussion (i.e. NNTP as a private discussion group transport protocol), the problem of one-to-none-via-many (nicely phrased :-) ) is not an issue. It's only an issue if you want to run a Usenet server (or a node in some other massive, public access, federated system) but we're not discussing that scenario at all. > If 25 years of advocacy haven't managed to get debian to understand > how wonderful it would be to switch from SMTP lists to NNTP groups, > it's really unlikely to happen going forward. Has there been any such advocacy? I wasn't actually aware of it, if so! My own participation in this sub-thread has simply been to point out that (a) NNTP still works as well as it always did, which is to say efficiently, elegantly and very practicably, for private discussion groups, and (b) that Usenet's problems related to its large scale/public federation issues do not apply at all to NNTP in the context of private discussion groups. All that said, I wouldn't want to see NNTP replace the existing email discussion lists. I personally prefer NNTP as an alternative, but not replacement, for email-based discussion. (As well as both email and NNTP as alternative access methods to web-based forums for those who prefer these seemingly outmoded but still flexible and efficient access methodologies). > Instead, it's likely that the theoretical advantages don't nearly > outweigh the practical disadvantages. The practical advantages of NNTP as a private discussion group medium are real enough, in my experience. :-) Footnote:- 1: Or, for that matter, receivers who came to their specific attention. The fact that receivers in Usenet over NNTP are less noticeable than senders is not something I see as a weakness of either Usenet as a whole or the NNTP protocol. The senders and receivers of DCC transfers in IRC, for example, also left no particular logging footprint at all. -- Mark Rousell
