Δαφνη Ευτυχια Αρθουρος

"In the sweetness of friendship let there be laughter, and sharing of pleasures." -- Kahlil Gibran

[And to my Jewish friends, happy Hanukkah (however you spell it).]

After being woken too early by an unexpected phone call, I checked my mail and then noticed that my shell account was nearing its disk quota. So I ran through my spam filter's "pretty sure this is trash" folder, saved one technically legitimate message that I haven't decided whether to train it to recognize as such (an ignoreable advertisement from eBay, where I do have an account; I should see whether I can modify mailfilter to do a three-way sort and put such messages into a separate "legitimate but reeeally low priority" folder -- I don't think I can concoct a Procmail rule to distinguish urgent eBay mail from eBay junk mail, but CRM114 could probably figure it out with enough samples), and checked my quota again. Then I thought about just how much space a day and a half worth of spam had taken up.

Thinking back to when I first started storing files on a hard disk and occasionally needing to back them up, archive them, or copy them from one computer to another[1], I realized that I had gotten ten diskettes worth[2] of garbage -- not counting messages that a separate filter upstream of CRM114 had identified as worms/viruses! -- in the last day and a half.

Of course, that's just one arbitrary and outdated scale by which to ponder spam magnitude. Even if I'd been getting that much junk mail back then, I would've been dialing in to a shell and deleting it on a remote computer (just as I did this morning using telnet instead), not copying it to floppies.

So here's another arbitrary and outdated scale: on the first computer I actually got mail delivered to -- uh, the first computer I sat in the same room with that got my email stored on it instead of my merely using it to run a terminal-emulator to log into some bigger machine elsewhere to read my mail -- it would've taken four hours to receive just my personal share of spam (and just the spam) via UUCP over our 2400 baud modem. Add in legitamate email, and it wouldn't have taken very many users[3] for there to have been more mail trying to get in that could physically be transmitted each day over the link we had. But of course, spammers would've had problems getting such huge volumes of mail out of their own computers and on its way, too. (It's worth reminding young'uns that the first spammers used Usenet, not email. That way a single copy of a message goes out via the spamer's modem and gets propogated to thousands of computers by rippling out (this computer tells two friends, and those computers tell two friends, and ... except that some computers had hundreds of 'friends') and the single copy that makes it to each destination computer gets read by several users there.

Really, news messages in Usenet propogated pretty much like worms do now, except that it was intentional, the messages only went to machines whose sysadmins had elected to receive them, and the messages didn't have to contain their own sneaky executables for sending themselves. (At a certain level, Usenet still works this way, but there are additional options for the "leaf nodes" now, in the days of ubiquitous TCP/IP connectivity.) So the first spam would have been nothing more than one rather annoying and inappropriate[4] but ordinary Usenet message, if it hadn't been repeated to hundreds of different newsgroups to make sure "everyone" saw it (which also meant that anyone who read more than one newsgroup saw the message multiple times

Ahh, but I remember the furor from a large number of users, the dire predictions, the "tragedy of the commons" arguments, and the explanations of why spam was a Bad Thing back then. And I remember the "just press 'delete' and ignore it. You're making a bigger deal out of it than it is," responses. And now I look at how many Usenet newsgroups got so swamped with spam that they became unuseable for the communities that had inhabited them, and how much money is spent by companies trying to keep spam from choking their mail servers, and messages missed because of misconfigured spam filters or overlooked because they didn't stand out strongly enough against the sea of spam in a busy user's inbox, ISPs and companies larger than a "soho" ('small or home office') having to deal with storing and efficiently accessing womdigious mail spools, anti-spam features at other folks' ISPs that interfere with my ability to send mail directly from my home machine (or any machine in a 'residential' IP address block), spammer and open-relay blacklists and the shouting when someone gets put on one by mistake or (at least they think) unfairly, friends who reject mail from entire continents, friends who have set up pain-in-the-ass "prove you're human before I let your mail through" traps (which, when working correctly, only fire the first time I send them mail from a new address, fortunately), people afraid to put their addresses on their own web pages, or doing so in mangled ways to foil harvesters, Usenet messages I can't reply to for the same reason (or have to reply to twice when the first attempt bounces because I didn't notive the "removethisifyouarenotaspammer" inserted into their return address), and folks still on dialup (until very recently, me) complaining about how long it takes to download mail that turns out to be mostly spam. And I still hear, though less often now, people saying, "What's the big deal? Click 'delete' and ignore it."

Well, if it weren't a big deal, there wouldn't be so much time, energy, and money spent struggling so hard to keep it down to sort-of-manageable levels, would there? We'd just be making jokes about it on late-night television and not buying industrial-strength mail-screening network appliances (unless you're a sysadmin or read the same magazines I do, you might not know about those), subscribing to third-party spam-filtering services, tweaking and tuning (and sometimes breaking, through oversealousness) software spam filters, seeing television commercials advertising touting an ISP's allegedly superior spam-blocking ... and still, after all that money and sysadmin-time has been spent, having end-users (myself and most of the people reading this) having to "just press 'delete'" as often as we do. I think most folks have a vague idea how much of their own time spam costs them; I suspect most have little clue how much money their ISP or their employer has already spent just getting the spam down to that level. If you're seeing less spam than I do, it's almost certainly because somebody upstream has already tried ver hard to delete most of it for you (so hope they didn't kill any legitimate messages by mistake -- that's why I switched ISPs by the way). Many of my friends get a lot more spam than I do.

When I started this with the musing that I'd gotten ten floppies worth of spam -- not counting viruses -- in the past day and a half, before I got distracted writing historical footnotes, I wondered: if everyone had listened to us back when spamming started, and had understood the magnitude of the problem we'd be facing now ... could we have done anything then to prevent it, or would it have just meant more people knew enough to be upset?

[1] That is, I'm not counting the machine where copying files meant going downstairs and sweet-talking the operator into making me a tape (spool, not cartridge), nor my first personal computers back when hard drives were expensive and rarely found in homes.

[2] Why, 9-sector 5.25" DSDD, of course. I think I only used single-sided -- or single-density -- diskettes on machines without hard drives, have only once ever used an 8" floppy disk (that's why 5.25" floppies are diskettes) and that was only to copy files off of it for a colonel, and I think I only used 8-sector DSDD (320K) floppies on a friend's parents' machine (running MS-DOS 1.0), if at all. And of course, quad-density and HD weren't around yet, nor 3.5" diskettes.

[3] That was a Xenix machine that I was sysadmin of at work. I think three people officially read thir email there, and another two or three of my friends did so unofficially. And yes, that was a very small number of users.

[4] Even after commercial advertising became kosher in some network spaces[5], repeating the message to lots of newsgroups guarantees that it'll be off-topic for most of them, if not all.

[5] Yes, there was a time when commercial advertising was universally frowned upon on the nets[6]. Back then the existing networks were for academic and military use, research, and, of course, the geeks' playground. They were a lot smaller then, of course, than today's Internet. There are still areas where advertising is frowned upon (please do remember that "the web" is merely one face of the Internet, one tool that uses the Internet), but it's not like the spammers give a damn about that any more than they care whether you want your mailbox stuffed.

[6] ARPAnet "grew up" into the Internet ... well, sortakinda -- that was the impression a lot of us got even while watching it happen, which is why that meme is so strong; really, ARPAnet inspired a bunch of similar nets, including NSFnet (the one that could more properly be said to gave "grown up into the Internet"), which talked to each other, and when the connections between them became so transparent that users couldn't really tell where one ended and another began, or which one they were on, it was the Internet. Hence the "inter" part of its name. But (to get to the point of this footnote) during the period I'm talking about, we also had BITnet and CSnet "real-time" networks (i.e., you could run a "chat" program on them) and probably others I didn't know about, which you could send email into or out of if you knew the trick, and Usenet (more formally UUCPnet, with Usenet merely being a conspicuous service that ran on it, much as the web uses the Internet) which was a "store and forward" network (email yes, chat no). That's why I felt I had to use the pkural, "nets", in the previous footnote. Me, I was a lowly UUCPnet peon, with a '!' in my name instead of an '@'. My cool friends were on real-time networks. (I was ..!seismo!dolqci!hqhomes!glenn)

[6 continued] The distinction between Usenet and UUCPnet became important, by the way, when ARPAnet users decided they wanted to be able to read Usenet ('news') messages. As soon as a few ARPAnet sites started gatewaying news, Usenet became a meta-network that used the ARPAnet for some of its links. So you were "on Usenet" if you could see Usenet news articles, but you might have been on ARPAnet, CSnet, BITnet, or UUCPnet as far as the mechanism of your connection to the world outside (and how Usenet got to you) was concerned. I'm not certain whether there are still any UUCP-only nodes connected; I would not be at all surprised but I expect them to be rare. I also expect to see the store-and-forward technology revived when we start trying to expand "the net" over distances where speed-of-light issues and bandwidth/power issues make low-latency or high-throughput impossible, or both. I'm talking about colonies on other planets or their moons, or a communications network between spacecraft scattered about the Solar system. I can see special-case applications for store-and-forward physical networks today, but not as a "lots of people use it for their email" thing (the way it was in the heyday of Usenet and FidoNet (a similar idea to Usenet that came along lateri (and, uh, was ... eventually absorbed into Usenet once both started using the Internet as a communications medium? I should research that...), built of personal computer Bulletin Board Systems (BBS) instead of minicomputers and mainframes). Store-and-forward logical networks, on the other hand are still with us but (mostly) using real-time networks for their physical layer. Email "relays", for example, if I understand them correctly. Note that I'm not counting writing email on your PDA or laptop while on the subway and offloading it once you get connected again as a store-and-forward network. The same goes for fetching your email from your ISP in a batch via POP/IMAP every so often. I'm reluctant to call a real-time network with store-and-forward endpoints for some of its individual users a "store-and-forward" network. An ad-hoc network that detected nearby Bluetooth devices and automatically handed your mail off with a request like, "please pass this on to someone closer to my destination ifwhen you get close enough," would be ... but I expect the spread of public WiFi hotspots and the various cellular-telephony broadband schemes to make the BluetoothNet (pocketnet?) not worth implementing except to show off.