NNTP servers were generally provided by your ISP, so ISPs built the capacity out as they added subscribers, and added groups based on the perceived interest. It wasn't especially difficult (and it's pretty trivial today) to offer all the "official" text-based newsgroups with a few days or weeks worth of retention; it only becomes a significant undertaking when you want to offer months of retention for binaries newsgroups.
Similarly, because the NNTP servers were basically a value-added ISP service, they were funded by subscriber fees. You never had a problem with a newsgroup suddenly getting popular and being unable to pay their own hosting fees, or suddenly tossing ads in order to pay the hosting bill, which is unfortunately common with webforums. And you don't have problems with a single server crashing and wiping out the history of a whole group. It's all pretty robust.
Architecturally, netnews was superior to web forums. Where it failed was in not having a good response to the spam problem. I think that had it not been for spam, people wouldn't have left NNTP for HTTP-hosted webforums nearly as quickly. But the decentralized nature of netnews meant there wasn't anyone who could quickly implement any of the myriad solutions that were proposed (cryptographic proof-of-work plus reputation systems probably would have cut down on the worst offenders).
My guess is that eventually, we'll reinvent something that's similar to NNTP but by another name (with the decentralization, server-server replication, gateways to other protocols, etc.), but it's been a long and unfortunate diversion in the meantime.
With web forums you fire a search term into a random crawler and it will likely barf up a url. With NNTP you had to configure the email client to point to the isp server, and then locate a relevant group for what you were interested in. And even then searching the whole archive was troublesome if not impossible.
I watched something similar play out with IRC. Once AJAX allowed random websites to offer chat services non-tech channels basically wasted away, as your random joe and jane just entered some url they were told, or read about in some glossy mag, and presto.
As for reinventing NNTP, i don't think there is any incentive any longer. Especially now that something as big as Reddit can just pay Amazon to scale the back end as large as they need.
Store and forward came to be before the net, and was used during the early net, because of narrow bandwidth.
And given that, you can't really have a store-and-forward edge-caching system that truly resembles NNTP. At most, you can get a hierarchical network like IRC.
Much as I distrust Google for anything these days, 8.8.8.8 / 8.8.4.4 offer a widespread advanced notice of domains tracking up the hit rates, and possible attacks.
Email, as STMP, after decades of nobel service really simply does deserve to die. It's far less about the mechanics of email user interfaces, for which there's a great deal of ink wasted, and much more on privacy -- encryption, metadata, and spam, probably as the three top concerns. Very nearly all the suggested replacements (SMS, various messaging systems, Facebook) have either those problems or those and the fatal condition of being proprietary silos.
The more I reflect on it the more the Web is close to what Usenet was, at least in some regards. Origins post messages, which may or may not die in short order (what's TTL for an Internet page now, a few months on average?). We're increasingly relying on sites such as the Internet Archive to actually provide a view-through-time at content.
What the web (and email, especially in mailing lists) lacks and has always lacked is a requirement to meet standards conformance. DDG "The Web is an error condition" and you'll find a far-too-true-to-be-good rant on that topic.
The loose-standards-anything-will-fly-nearly approach has allowed a great deal of flexibility in page design, but much of that is ultimately painful. I'm finding a much greater appreciation for straight-up HTML 1.0 formatted pages.
A huge part of the problem is exceptionally crappy entity styling defaults. If browsers presented pages sanely (at various resolutions), much of the B-Arc profession of Web design would simply cease to exist. And this would be a wonderful thing.
A simplified markup, acceptance criteria (and rejections for failed formats), cache-and-foreard, client-determined-formatting, security, and either a universal content syndication or UBI type compensation model might address other issues.
I'm getting the sense of some interesting possiblities out there.
Actually, when I DDG this phrase, the only result is your comment in this thread.
> One day, I walked into the break room and heard a coworker say, “The Web is an error condition,” referring to the deplorable state of code out on the Web. I think that was the end of the end for me, because it just depressed me. It depressed me not because it was untrue, but because it was so perfectly true.
> Honestly, I miss the days when Netscape Navigator would just halt rendering in the middle of your page, saying, “No, I will not parse any more of your shit until you fix it.”
> Then IE came out for free. Suddenly, the game of web browsers changed from paid apps to supported by advertising and search revenues. The only way to get users to use your browser (and thus get more money to develop with) was to parse all the shit you used to reject.
> The web became a co-evolution of crap and trying to render crap. (It’s gotten more complicated since then, but because there’s been a habit of rendering crap, no one suddenly wants to stop.)
The Android is also an error condition. Trying to copy/paste and maintain state whilst composing anything complex generally fails.
Yeah, but think of the social scaling issues. You simply can't have every single person interested in e.g. American football reading & commenting on alt.games.football.
News was indeed a far superior architecture & experience to web forums — so long as the userbase was relatively small.
Gene "Spaff" Spafford thinks 50k - 500k users ~1988 - 1994 is probably ballpark accurate. Peter da Silva notes that in 1990 he was the most active user on Usenet ... and under another handle, the 3rd most active. Microsoft did some early 2000s research turning up low-single-digit million distinct user IDs IIRC (I've posted on this previously).
So, yeah, it was small.
Organising people into groups, even ad-hoc, from which exceptionally good messages might be shared to others, might be one way of handling scaling. Clustersizes ranging from a total size within Dunbar ranges, to a 1/9/90 split giving the 10% active users a Dunbar cap (about 300 - 3,000 users per cluster) might work.
I agree that 1990 Usenet in 2016 would be a fatal mistake.
Upvoting. NNTP didn't have it. Reddit does. This allows a reader to see the top posts.
for a while and I don't recall getting any spam, ever.
I think Gmane 2.0, if it ever will be born, would have a mechanism of plugins, that allow easy wrapping any web-forum in it, by writing a little template describing the layout of pages like 'this is topic, this is message body' and so on. And the crawler to use the plugins, of course.
Well, except for the mess that is RFC 822. At least NNTP decided to go 8-bit safe eons ago, so you just have to shove everything in UTF-8 and you can screw RFC 2047 in the face.
* RFC 2047 is a pain in the butt
* Handling attachments can be difficult (particularly the inability to safely add binary data)
* The display name/address is both painful in terms of syntax.
* CFWS in general (you can tell who's written an email client by their reaction to the letters CFWS)
* Header fields in general have way too lenient handling, so there's no way to, e.g., guarantee non-ASCII handling (RFC 2047 needs to be handled differently in the Subject: header than from the From: header, and the Content-Type: header does something else entirely).
* Mandatory line-length limits, which means you need to be able to insert line breaks in however you encode custom header values
* No way to get attachments without the body or vice versa
* All metadata needs to be shoved into a header (see above about i18n, line length, and CFWS)
What would you add? Remove?
Why?
For myself: the basic RF282 concept of sender, subject, recipient (or newsgroup), date, and references, plus body, seem reasonable.
How best to take these to present capabilities (and needs) is another question.
I suppose that it's one more nail in the coffin of NNTP. I really miss net news, but it probably wasn't possible for it to scale to support the entire world. It's sad, though: for a short time, all the interesting people in the world were able to communicate with one another.