Joe Dellinger
This article appeared in the September, 1992, edition of ``The Leading Edge'' on pages 37 to 40.
© 1992 Joe Dellinger and the SEG
Electronic mail (e-mail for short) was originally a way for computer users to send message files to other users on the same system. E-mail's usefulness has greatly increased today because so many of the world's computers are connected to networks. A network can be as humble as one between two machines in a room, or as big as a globe-spanning behemoth linking tens of thousands of machines at US military bases around the world. Whatever their size, more and more local networks are now interconnected with other networks. The link can be as inexpensive as the UNIX program UUCP, a simple-minded automatic program that uses a modem to call up other computers on the telephone, or as fancy as special-purpose gateway hardware linking multiple fiber-optic networks. In any case, gateways route information from one network to another while performing the necessary security checks and format conversions. Because the world's networks are so interconnected, it has become possible to send e-mail virtually from anywhere to anywhere.
If you currently work for an oil company or university in North America and have a login on almost any sort of ``locally'' networked computer system, chances are you can already exchange e-mail with associates all over the world (even if you don't realize it yet). If you are one of the many people who could use e-mail but don't, perhaps now is the time to start!
When you give your telephone number to someone in another state you don't tell them ``just call extension 7822''; you know to give them a complete number, including an area code, which will work from any telephone in the US or Canada. In an ideal world e-mail addresses would be as standard and well understood as phone numbers. If someone tells you what they think their e-mail address is, you should be able to send mail to that address and the message should speedily get to them. This should be true regardless of what kind of machine you have, what sort of machine they have, and what networks and gateways happen to lie in between.
E-mail is not yet as easy and reliable as the telephone. Many different address formats are still in use, and the gateway programs interfacing local networks to the rest of the world are often locally produced efforts that harbor many idiosyncrasies and outright bugs. It is already quite clear what address format will become the global standard, however, and that is internet-style addressing (which is especially convenient for you if your site is one of the rapidly swelling hoard that is directly connected to the internet in some way).
My workstation is directly on the internet, and my (internet-style) e-mail address is:
joe@montebello.soest.hawaii.edu
(Note that on most systems it is possible to create a short ``alias'' for commonly used e-mail addresses, so you don't have to remember and type such long strings of gibberish every time you want to send mail. This is especially useful because some sites seem to enjoy using random strings of numbers and letters for their user IDs and workstation names.)
Internet addresses are a bit like US postal addresses in that they
begin with the local and work out to the progressively more encompassing. My
address indicates my user name is ``joe
'' on a workstation named
``montebello
''. The local subdomain containing montebello
is ``soest
'',
which is named for the School of Ocean and Earth Science and Technology
it covers. This subdomain is in turn contained within a domain called
``hawaii
'', which
is the domain for the University of Hawaii network.
Finally, this is all contained within the root-level
domain ``edu
'',
which joins together universities throughout the United States. Users
within SOEST can send me e-mail using just
the address ``joe@montebello
''
if they prefer, but only the full name will work out for people outside
the ``soest.hawaii.edu
'' domain. The full name
``joe@montebello.soest.hawaii.edu
'' should always
work for anybody in
the world whose machine is connected to the internet. Such ``standard''
addresses also work at an ever-growing number of non-internet sites as
well, because more and more gateway software knows how to translate
internet-style addresses into whatever format the local network demands.
Note that the layered design of internet addresses means that your computer does not have to ``personally'' know of every machine you can send mail to. If it doesn't know how to send the mail somewhere it can ask a local machine called a ``name server'' for help. If the name server doesn't know either, it works down the address looking at progressively bigger and bigger domains until it finds one it knows of, and in turn asks for help from there. This decentralization is so complete that even the people who run the internet don't know how many machines are on it!
For example, suppose you are trying to send mail to my address above.
Your workstation probably won't know of mine, so it will have to ask a program
called a ``name server''. In the worst case the name server won't recognize
any of those domains, and will have to ask a machine responsible for root-level
domains for a list of machines that can be asked about ``.edu
''.
The name
server will then query these machines about ``.hawaii.edu
''.
Machines
knowledgeable about ``.hawaii.edu
'' will in turn be asked about
``.soest.hawaii.edu
'', and so on until
the name server finally obtains
a direct address for montebello to give to your machine. After this furious
second or two of activity, your machine can directly connect to mine and
deliver the mail. It can also happen that the search will end at some sort
of security gateway, which will refuse to provide further access but will
officially accept mail for forwarding on to the final destination. In this
case your workstation will simply hand the mail over to the gateway for
forwarding.
This last mechanism is quite general and comes in handy for more than just security firewalls. What if you are not on the internet yourself - what address can you give out to the rest of the world? Through the magic of ``MX (Mail eXchange) records'', non-internet sites can register an internet-style address with the network authorities. (In the United States this is quite inexpensive.) The network will then know how to automatically and invisibly send e-mail to that site by forwarding through the required gateway. For example, when I send mail to
``jim@killer.dallas.tx.us
''
I don't need to know that this e-mail will shoot over the internet to UT Austin in a flash, and then slowly wend its way via phone calls through several privately owned small machines before reaching its final destination on a Dallas BBS (bulletin board system). MX records make it irrelevant to outside e-mail users whether a registered site is physically on the internet or not.
In these wild-west days of networking you shouldn't expect perfection.
(I'm sorry to tell you that most Geophysicists are not exactly at the vanguard
of network communication.) I get a fair number of e-mail messages with bogus
reply addresses from Geophysical sites because of faulty local gateway software
that knows how to send e-mail on out to the wide world but doesn't bother to
transform the local
``Reply-To:
''
address into something generally useful. (This is
the e-mail equivalent of the annoying people who leave urgent telephone
messages without telling you how to call them back.) In their attempts to
reconfigure their e-mail software several sites I've known have accidentally
instructed the network to forward their e-mail round and round in circles,
or done something else equally fatal so that all incoming mail is lost or
rejected for a few hours (or days) until the problem is noticed and fixed.
(It isn't the system administrators' fault that these sorts of software
problems are so common. Judging by the ``ease'' with which UNIX sendmail
configuration files can be understood, software vendors evidently think
that e-mail configuration is a sport for system administrators who might
otherwise be forced to seek their mental entertainment from deciphering
ancient Etruscan monumental inscriptions. Hopefully the situation will
eventually improve.)
There are still quite a few geophysical sites around who require their users to jump through hoops to send e-mail outside their local network. For example, researchers at a well known site in Britain have to use the address
CBC%UK.AC.NSFNET-RELAY::EDU.HAWAII.SOEST.MONTEBELLO::JOE
to send me e-mail. As you can see, in Britain the nationally mandated
e-mail address standard reads in the ``wrong'' direction (compared to
the Internet). This is the result of a ``standards war'' Britain lost,
somewhat reminiscent of the one over which side of the road to drive on
in Europe. The long address above manually routes the mail out of the local
network to the British academic network JANET, from there on out to an
Internet gateway, and thence to me. The address is complicated because it
explicitly mixes several distinct styles of address. Fortunately the JANET
domain-order reversal is hidden from the world outside by smart gateways at
the British network frontiers. When the mail sent on its way using the above
address reaches me it appears to have come from a standard Internet-style
address. When I reply to that standard-format address my mail makes the
complicated return trip without getting lost. (No doubt within a few years
Britain will be 100% standardized, and anyone there will be able to reach
me as ``joe@edu.hawaii.soest.montebello
''!)
If internet-style addresses don't yet work at your site or somebody
gives you a non-internet-style address to send e-mail to, look at the AGU
membership directory for ideas to try. Many sorts of address are ``almost''
internet-standard, because while not quite in the official nested-domain
format they will work on most systems. Such addresses are usually of the form
``a%b.c@d
'', where ``d
'' is the address
of an internet machine that
understands the non-internet address ``a%b
''.
The ``.c
'' lets the gateway
machine ``d
'' know what
non-internet network this e-mail needs to be routed
onto.
For example, when I send mail to Ivan Psencik in Czechoslovakia I use the address
IP%CSPGIG11.BITNET@IBM.RZ.TU-Clausthal.DE
My machine just ignores all the stuff before the ``@
'' and
sends the
e-mail to a machine in Clausthal, Germany, with instructions to deliver
it to the user ``IP%CSPGIG11.BITNET
'' there.
The machine in Germany changes
the ``%
'' into an ``@
'', notes
the ``.BITNET
'' ending, and in turn forwards
the mail on to Czechoslovakia using the BITNET
address ``IP@CSPGIG11
''.
(Exactly one ``@
'' at a time in an address is allowed.)
In my opinion it will be more productive in the long run for everyone to insist that this sort of headache should be solved by gateways and not users. If your site does not have a standard internet-style e-mail address that works from the outside world, bug your system administrator to do something about it!
There is one big problem that I haven't yet covered: as the Russians
found out, telephones lose much of their commercial usefulness if telephone
directories are not widely available. While there is an internet-wide e-mail
directory system, it is not much use for geophysicists.
``whois dellinger
''
finds the e-mail address for someone working
at ``USAHC-PENTAGON
'', not me.
(The internet was originally a military network intended to link US military
bases with research institutions, and the official directory mostly only lists
military types.)
The American Geophysical Union compiles e-mail addresses in their
membership directory, although these are not yet available on-line. The
SEG does not yet compile e-mail addresses at all, but I expect it will start
doing so soon. In the future I expect professional societies like the SEG
will routinely facilitate e-mail between users by providing on-line
directories. Then anyone could find the address, say, of Sven Treitel by
sending a message like ``address Treitel
'' to an
automatic directory program
at ``directory@seg.org
''.
(This sort of process will probably eventually
become standardized through the currently experimental decentralized
``white pages'' directory system.) It should also be possible to provide
forwarding services; for example,
mail to ``ken.larner@seg.org
'' could
automatically be forwarded to Ken Larner's current e-mail address. (This
facility would have to be optional; many researchers treat their e-mail
addresses like unlisted phone numbers, to be given out only to a select
group.)
Another way to facilitate communication is via mailing lists. For an example, I maintain a mailing list of people interested in seismic anisotropy. I started the list by simply announcing its existence at SEG sessions on anisotropy between talks. Once it grew past a few dozen names, it became well enough known that it now mostly grows by people sending me e-mail and asking to join. The list can be used to send messages to everyone on it. This ``broadcast'' facility is very useful for conference announcements and such general information. While it is also used now and again for brief bursts of technical discussions, it isn't used as much as it could be because of the reluctance of researchers to see their results widely and instantly disseminated to large numbers of possible competitors.
Mostly the list serves as a directory of e-mail addresses for researchers working in the subject, and the communications facilitated by the list are mostly private. That is useful enough. Because of the list, I find I can often directly correspond with the author of an article that I find interesting, something I simply couldn't have conveniently done otherwise. If you want a copy of the list for yourself, send e-mail to
``listproc@sep.stanford.edu
'' with the body
``recipients anisotropists
''.
(If you don't get any answer your reply address is broken!)
After reading the previous sections, many of you have probably come to the conclusion that letters, express mail, telephone calls, and faxes are all more reliable, easier to understand, and go more places. Why bother with this less certain e-mail thing? It is certainly true that the traditional communications media have their place, but it is also true that e-mail makes certain types of communications possible that just weren't possible or convenient before. Most young academic researchers I know consider e-mail as an indispensable tool for their trade. Why?
For one thing, E-mail is cheap and convenient for quick questions, in contrast to long-distance or international telephone calls that can be prohibitively expensive and are certainly disruptive to the person's schedule at the other end (if you can even catch them in their office because of the time-zone difference). On most e-mail networks you will get your answer within minutes of the other person's reply.
Consider how the University of Hawaii uses e-mail to keep in contact with their research ship, the Moana Wave, at a reasonable cost. Researchers on the ship can forward their campus e-mail to their ship accounts if they wish, so their normal shore addresses continue to function even while they are away. Similarly, the researchers at sea can send e-mail to colleagues on shore just as easily as they could when on land. The queued messages in both directions zip through the satellite telephone connection once a day in just a few minutes. This is a much more efficient and convenient use of a ten dollars per connection, ten dollars a minute international-monopoly satellite telephone call than the captain on the ship forwarding messages by reading them out loud over the phone and then scribbling down the messages read back.
If you are jointly editing or writing a document with a distant associate, e-mail is useful because it allows you to send back and forth the electronic source for your document straight from the machine you are using to edit it to your coauthor's machine. This is certainly better than FAXing back and forth marked-up paper copies which have to be re-entered into the word processor on each end again and again. The advantage is even more apparent when you need to send a copy of a computer program to someone else. The only real alternative to e-mail is mailing a tape, which takes longer and is certainly less convenient (not to mention the difficulties of getting tapes written by someone else's drive to read).
Convex computer corporation prefers to receive and respond to bug reports via e-mail; this ensures a written record for both sides to refer back to. At sites with commonplace workstations susceptible to breakins because of their cookie-cutter ubiquity, your system administrator can receive the latest cracker alerts and security hole information via e-mail, and then can get appropriate security patches and information the instant they become available. If you are stymied by a bug in Oracle's database software and need a quick fix, Oracle's support people will probably ask you to e-mail them the offending program, and then e-mail back the fix. If you don't have e-mail, you have to wait at least two days longer and pay Federal Express for the privilege of overnight delivery both ways.
With the advent of page-description languages like Postscript,
it is now possible to e-mail typeset documents, even ones including figures.
This means it is possible to submit documents such as AGU abstracts, NSF
proposals, or even entire journal articles electronically. (Some journals
already do this, and in the future it is quite likely that the primary
means of distributing journals will be via e-mail.) Software is currently
available to support an extended form of e-mail that embraces text, images,
and sounds. At least one consultant I know already uses e-mail to receive
and send SEGY data files. As computers and networks get yet more powerful
constraints on message size will ease, and I expect new standards to
revolutionize how science is done. Using e-mail you will be able to retrieve
an article or even an entire book from an archive anywhere in the world at
nominal cost. If the original author wished to make their results verifiable,
the message will contain not just the article, but the data, the algorithms,
and all the steps required to reproduce the results. (Jon Claerbout's
``e-book'' ``Earth soundings analysis: processing versus inversion'' may be
the first example; his address is ``jon@sep.stanford.edu
'' if
you are interested in how he did it.)
The internet is already huge and the number of sites on it is currently estimated to be increasing at the rate of about 5% per MONTH. The global e-mail network has already reached Norway, New Zealand, Novosibirsk, and even Namibia. Should you manage to find out their semi-secret addresses, you can even send e-mail to Antarctica and outer space (the Russian space station Mir has packet radio). E-mail is certain to be one of the most important communications tools of the future. Why not get ahead of the game and start making use of it now?
Footnote: The e-mail addresses for Jon Claerbout, Ivan Psencik, the anisotropy mailing list, and me are correct and ought to work. Thanks to Jon and Ivan for giving permission to use their addresses as examples. The other addresses, used only as examples without specific names, are ``slightly modified to protect the innocent'', but are nevertheless authentic in spirit.