I've posted it a little early because I'm excited but here is my reading and analysis of RFC-1, the very first official Request for Comments document and an important piece of internet history.


I'm doing one of these a day for a year. You can follow along at @365-rfcs. I'll only be posting the really noteworthy ones on this account.

RFC-5 was an early (eventually abandoned) proposal for delivering rich applications over ARPANET. Specifically it was conceived as a way to connect to Doug Englebart's "mother of all demos" computing system remotely from a more typical OS! My writeup:


You can follow along with my series commenting on the first 365 RFCs here @365-rfcs

RFC-7 led me down an unexpected computer history rabbit hole where I learned about GORDO, an operating system that was quickly renamed to... SEX. Yeah.


You can follow along with my series commenting on the first 365 RFCs here @365-rfcs

RFC-15 is an important one: it's the initial proposal for Telnet! The first version of this program was written in late 1969 and it's a tool that I still occasionally use today, which is really amazing when you think about it.


You can follow along with my series commenting on the first 365 RFCs here @365-rfcs


Okay, RFC-20 is here! This is the RFC that says "we are going to use ASCII for communicating between computers". Read on to learn about what a character encoding even IS and why it still affects our day to day internet experience. It'll be a very%20fun%20read, I promise.


You can follow along with my series commenting on the first 365 RFCs here @365-rfcs

I contacted the Computer History Museum and paid them a small fee to have them scan the first 9 RFCs. I'm happy to say those scans are now online.

My post with interesting excerpts and things I learned looking at the scans: write.as/365-rfcs/update-scans

The listing of the scans in their catalog, with a link to the PDF: computerhistory.org/collection

You can follow along with my series commenting on the first 365 RFCs here @365-rfcs

Hello from the storage warehouse of the Computer History Museum, BTW.

In my comments on RFC-32 I go on a digression about how much I love the 555 timer integrated circuit chip. Also, computer clock nerds, please check my work and make sure I did the right calculations and estimates. I'm not so fluent in this stuff.


You can follow along with my series commenting on the first 365 RFCs here @365-rfcs

Okay, so: based on my research I believe that RFC-69, from September 1970, is literally the first example of an internet person doing a reply-all and saying "please unsubscribe me from your list".

This is, of course, nice.


RFC-84 is an interesting one: the first RFC to categorize RFCs. This is authored by Jeanne North / Reddy Dively, who was one of the many women librarians who made the computer revolution possible. I write a little bit about her here. write.as/365-rfcs/rfc-84

@darius you were not kidding about the hand-drawn diagrams being great

@darius Interesting! Were the RFCs previously in some kind of online system (maybe NLS itself?) and then lost to the world of computation until now? Or were the original copies actually made of paper?

@kragen They were mostly typewritten and copied and mailed around. Some were hand written! They were transcribed into text form in the late 90s so before then they were indeed not online.

@darius Surely 10E7 is not a hex value, but equivalent to 10e+7 = 10*10^7 = 100 billion? Although “1 part per 100 billion” error sounds too good, then.

@ryantouk That would not be the correct accuracy for a relatively inexpensive crystal clock, yeah.

@darius WAIT 10e+7 is 100 million, not billion, which is now 1000x more feasible. Oops!

@ryantouk yup, still it means 1 ms of clock drift a day which seems... overly ambitious

@darius 10e7 seems likely to be scientific notation: e7 == 10^7.

@darius kinda ambiguous, so I would tend to look for what the authors used in other documents. If they used scientific notation a lot, then that tips it. (I love this project so much, btw!)

@darius great read.

my arduino loses dozens of seconds per day so 1ms seems crazy for 1970. so i was googling around for this odd accuracy thing and i think it's not a percentage of error per day but the frequency error. but then i'm not sure how to interpret this either. found this ham radio blog using the same terminology https://www.febo.com/pages/stability/

@oberhamsi @darius Commercially-available TXCOs could do +/-5 pp 10^7 stability over a very wide temperature range at that point. That was available by the mid- to late-sixties. 1 pp 10^7 for an indoor stationary mount wouldn't have been out of the realm of normal.

@jond @oberhamsi @darius Yeah, I find that believable, after some research. If not a TCXO, then an OCXO for sure. Though I am not sure what can be considered "low cost" at the time.

I am also of the opinion that the 10E7 thing is a typo, and it is indeed meant to be 1E7, since 10E7 is neither normalized nor is the exponent aligned to a multiple of 3. I have personally made typos like that.

@jond @oberhamsi @darius Also, sorry for the lingo.

TCXO = Temperature Compensated Crystal Oscillator
OCXO = Oven Controlled Crystal Oscillator

@dratini0 @jond @oberhamsi @darius I thought that was strange too. Would have expected 1E8 or 100E6.

@darius I would really like to go back in time and tell that person that in nearly 50 years' time, someone will explain their unsubscribe request to the internet at large (which is a thing now), and will also make a sex joke about it.

@darius That's amazing. So the next question is, did that actually get distributed or did they just assign it a number and file it, later to be dug back out when they put them online?

@freakazoid It was copied and distributed. I've handled one of the originals personally.

Sign in to participate in the conversation
Friend Camp

Hometown is adapted from Mastodon, a decentralized social network with no ads, no corporate surveillance, and ethical design.