AIM INTERVIEWS

Who are you?

Eric Allman, formerly a student at UC Berkeley during the early UNIX days and later a staff member at U.C. Berkeley, Britton-Lee, the International Computer Science Institute, and co-founder of Sendmail, Inc. Best known as the author of sendmail; also the original author of syslog (which has been re-written), the –me troff macro package, and a number of small UNIX system facilities. I was also Lead Programmer on one of the first relational database systems (INGRES) in my student days. In my career I've worked on neural networks (same as used in deep learning today, but we didn't have the hardware to make it sing back then), infrastructure support, a streaming SQL server and a research prototype of a secure distributed storage infrastructure. I play with IoT things at home.

Why did you write the original ARPANET delivermail? Was there any inspiration?

The INGRES project had a PDP-11/70 with a connection to the ARPAnet. Everyone in the CS department wanted access, but we only had two spare terminal ports (ports were expensive back then) and the local network (BerkNET) was a store-and-forward system (i.e., you could do slow file copies and send email, but no remote login, no web style interface, nothing interactive). I found myself in the middle of arguments about who got access to those two ports, but as a student I really didn't want to mediate in arguments that involved faculty members.

Fortunately, at some point I realized that what most people wanted access to was not the ARPAnet per se, but email. I realized that with a little bit of ingenuity I could write software that would forward email between networks, even those that had wildly different protocols. Thus, someone on a machine with BerkNET access could get their mail through using painful syntax like "ing70:user@mit-xx". That software became delivermail.

Delivermail was a quick hack though, and had a lot of problems, especially as it got distributed beyond Berkeley and places wanted to support other networks. Most importantly, it didn't support SMTP, the mail protocol for the still-in-development Internet. This required substantial code changes, and there is very little left of delivermail in sendmail.

You created sendmail in 1981 and over the next decade it routed the majority of email on the internet. Why did you later found sendmail, inc in 1998?

SMTP already existed and sendmail already was the de facto SMTP for the Internet, which was part of the problem. As my user base got into the thousands of sites, my time was consumed more by support than development. Remember that I still had a day job at that point, so sendmail was a nights-and-weekends project. In short, I wanted to get back to coding (i.e., the fun part), and so my original thought was that the company would be primarily a support organization. I was convinced by a friend of mine that to start a real company I needed loftier goals than that. My friend became my co-founder. Atypically, we are still friends.

If you had unlimited resources (time, money, etc.), how would you spend your time?

If I had truly absurd amounts of money, I would probably be a MacKenzie Scott style philanthropist. But at some point money becomes a burden, so I'm glad to be of more modest means. Since I am retired I (in theory) have plenty of time, but it does seem like there are always household chores to do. But I wouldn't like having an on-site staff to take care of these things, so unless days can be made longer or I can convince myself that I don't need to sleep, that's probably a non-starter. I already travel quite a lot, see friends frequently, and am blessedly free of any serious health issues. Basically, I'm not sure I would change a lot in my life.

What's a message you have for the world?

Don't listen to other people telling you how to live your life. Do listen to as many different ideas as you can handle, but don't believe all of them.