Help me with my e-mail ponderings...

January 22, 2007 9:42 AM

This is a two-part question concerning e-mail, the internet's original 'killer app'.

i ) We know email has its flaws - but I'm looking for reasons (obvious and non-obvious) why an email address directory for the entire net has never (been attempted?) worked.

ii ) Does anyone know of a web app (preferably AJAX) which would serve as a repository for a given group's email addresses? Sort of a smaller version of the above. Individuals would be able to update their record in the app (perhaps by verifying with the system they own the address), there would be sensible authentication processes and vCards could be generated on the fly? It strikes me this could be quite handy...!

posted by dance to computers & internet (16 answers total)

(1) The internet is Really Big and grows dynamically. I own around a dozen domains for personal use, and can add new email addresses at any time. Google's GMail probably adds thousands of email addresses a day.

Spammers and bulk emailers are both aggressive about attempting to complete this project. Yes, it is useful for them. No, I don't want to be on the list.

(2) Most web based email platforms do this already. Take a look at GMail. Corporations tend to like products like Outlook+Exchange, as it gives them more flexibility in managing address books.
posted by b1tr0t at 9:46 AM on January 22, 2007

1. Way too big, spam, constantly changing, privacy, etc.

2. Plaxo
posted by stovenator at 9:48 AM on January 22, 2007

i) I realise the net is huge and I realise there are billions of email addresses - it just seems perverse that you either know a person's email address and can contact them, or you don't, and can't...

ii) Perhaps I didn't phrase this well, but neither Gmail nor Plaxo seem to offer the functionality I'm looking for - this would be a web app for an entire group to use, not an individual to use to keep track of a group.
posted by dance at 10:17 AM on January 22, 2007

I can't comment on 2, but I'll take a whack at #1.

To my knowledge it's never been seriously attempted. I generally try to avoid blanket generalizations, but I'll go out on a limb and say that it will probably never be attempted, and it certainly won't succeed with the email system we have today, for several reasons:

1) I would say the email system was designed to make this impossible, but applying the adjective 'designed' to email would stretch credibility. What we call 'email' is a combination of a few different protocols that were written out of necessity to enable users in wildly different networks to communicate. In a very real sense, email predates the internet as we know it. As such, the actual format of an email is governed by a hugely permissive specification that makes even validating an email for correctness a tricky/impossible task depending on your definition of 'correct'.

2) Email has no security, in fact email is anti-security. If a site were to publish or allow to be easily discovered a list of valid, active users, that would make blanket attacks substantially more effective, as attackers would not need to guess usernames in addition to passwords; they already have a list. For this reason, many administrators disable no-such-user bounce messages to seal this possible information leak.

3) Administrators can, do, and always will deviate from the spec of what email should be if they feel they must. The aforementioned not-bouncing is only one example, there are thousands. Even if all mail software were capable of adhering to a standard (*coughexchangecough*), it would be configured by admins to modify that spec if they felt it necessary.

4) Size. Without quoting Douglas Adams, the possible address space in the email system is staggering. A trivial case, of four-character domain names plus three different extensions (.com, .net, .org) and eight-character usernames yields ~ 14,215,144,014,964,850,688 possible email addresses. That's very much a contrived example, but you get the idea. For comparison, 1.4215144 × 10^19 is in the ballpark as the number of grains of sand on the planet (7.5 x 10^18)

5) Economics. There's no compelling use case for such a beast, except for unsolicited email. Nobody wants to receive more spam.
posted by Skorgu at 10:21 AM on January 22, 2007

Just to flesh out that size example, assuming a beefy server could poll 600 email addresses per second, that contrived example would take >750 million years. Even assuming Google's server farm (say 50,000 servers, 600rq/sec each) it'd still take over 15,000 years. And that's very much a miniscule subset of all possible email addresses worldwide.
posted by Skorgu at 10:25 AM on January 22, 2007

There was indeed a paper internet "telephone book" in the early days. 1992? 1993? It listed email addresses as one would list phone numbers in a phone book. Of course, it became useless when the internet exploded.
posted by YoungAmerican at 10:30 AM on January 22, 2007

Here's the problem. Even in my trivially small personal e-mail address book that I manage (maybe a few hundred entries), addresses change too fast for me to keep up with.

Now, scale that up by a few billion people, especially where a single person might have dozens of valid "base" addresses, and a system that continually generates new short-term addresses from these base addresses -- "short term" meaning that they might only be valid for one day, or maybe even just a few hours.

This kind of task is fundamentally impossible to scale. It's difficult enough for a single company to manage all of it's own internal addresses.

You might as well try to re-arrange all the atoms in the entire Universe before you try to solve the bigger problem of trying to keep a perfect database of all possible e-mail addresses.
posted by bradknowles at 10:50 AM on January 22, 2007

I actually recall some email directories on the web around '95 or '96. But of course they soon ran afoul of the problems that others in this thread have already outlined.

It occurs to me that a useful directory of email addresses simply cannot exist until some form of government regulation is in place to deal with the attendant problems. I mean, think of all the regulation that goes along with the phone network. Yeah. We're talking something on that order. If phone service throughout the 20th century had been as decentralized and chaotic as email service, would we have ever had phone books or directory assistance the way we do now? It's pretty doubtful.

Heck, look at the cell phone revolution. In many ways, it's been kind of a microcosm of the land phone phenomenon: advanced, expensive communication technology gradually become more affordable and more ubiquitious. But there were some important differences too. Cell phone networks have never been as centrally controlled or as regulated as land line networks. And has anyone even seriously suggested a centralized directory of cell phone numbers? I've never heard of any such proposal. It would just be impossible to carry out.

However, there is a case to be made here that once email becomes a necessity to the same extent that phones have, we will indeed see the necessary level of regulation and, with it, some sort of centralized directory system. God only knows how long that will take.
posted by Clay201 at 11:07 AM on January 22, 2007

Phone directories are possible because there are only a few bodies authorised to give out phone numbers, and collating the information from them is relatively simple.

Anyone with a computer and an internet connection can give out e-mail addresses, and they don't need to register with anyone to do so. This makes collating the information extremely difficult, putting aside the very real problems of scale.
posted by bonaldi at 11:13 AM on January 22, 2007

Even if someone wanted to make a listing of e-mail addresses and who uses them, how would they do it?

It's conceivable that in some circumstances it would be possible to have such a relationship exist, but e-mail doesn't work the same way physical addresses do. One physical address, one physical space.

Here at work, we have one e-mail address that never gets used to send mail, but receives mail and delivers it to ten different people. How would you classify this one?

My personal e-mail is filtered through several means. It's even possible for me to use something like TDMA to have almost infinite addresses for personal use, without much work. How would you classify each of these?

Lastly, and probably most importantly, what would be the point?
posted by odinsdream at 11:32 AM on January 22, 2007

There's also the constant problem of disposable cell phones and email addresses, a problem you just don't have with landlines.
posted by jedrek at 11:39 AM on January 22, 2007

Infospace.com still has a (lame) email lookup. There used to be others as well.
posted by loiseau at 12:30 PM on January 22, 2007

An email directory just doesn't make any sense at all. Putting aside all the reasons already given, if someone wants to be contactable by random strangers they make a web site that tells how to contact them. If you want someone's address you go to their page and find their contact information. Google works quite well here. If you can't find such a contact page, they don't want to be contacted by random strangers; this being the case for probably 95% of email addresses in existance. Trying to make a directory is just trying to solve a problem that doesn't exist -- the people that want to be contacted by strangers already are (via google), and the people that want privacy wouldn't participate in any kind of directory system even if it did exist.
posted by Rhomboid at 12:34 PM on January 22, 2007

At an organizational level, companies use LDAP servers and clients for directory services. Many of these, including Microsoft's ubiquitous Exchange software, are usable over the web, although I'm not aware of one that uses Ajax.
posted by ikkyu2 at 1:22 PM on January 22, 2007

Internally companies everything from LDAP to paper directories, stopping through custom applications, internal wikis, intranet pages and probably shared text files on a fileserver if you looked hard enough. None of this is (or should be) accessible to Joe Internet User.
posted by Skorgu at 2:13 PM on January 22, 2007

LinkedIn may go some way to what you want. It's not so much an email-address directory but Yet Another Social Network, which (soc nets, not LinkedIn) are probably the only way we have currently for the dissemination of email addresses that doesn't feed spammers.

Consider it this way: you want people to be able to find your address... but only people who already know you or for whom some trust relationship can be established.

Retrofitting trust into email is difficult, but we can enforce it in how we distribute our addresses.
posted by polyglot at 8:12 PM on January 22, 2007