Originally published on posts.postlight.com on February 22, 2016. Thoughts on an old bot and a column written by a new one. It’s the Postlight Newsletter for Monday, February 22, 2016 — Subscribe here.
I know lots of bot folks (I know that sounds terrible) and for some reason none of them seem to know about Racter, an “experiment” in “artificial insanity” that was at its essence just a particularly weird chatbot; it spewed nonsense.
I remember we got our copy cheap at a store in Downingtown, Pennsylvania, in 1987, for maybe $15. Consider: In the 1980s people would go to a store, buy a commercial chatbot off a shelf, take it home, remove the shrinkwrap, skim the manual, put the disk into the drive (chunka chunk) and type at it.
It was a different time. We didn’t really know what computers were for back then. Sure, spreadsheets, and word processing, and games, and then…whatever sold?
Racter (apparently short for “raconteur” but also, hmm, the latter part of “character”) was a vertically integrated chatbot; it was available on multiple platforms, the language used to create it (INRAC) was for sale to other bot-makers, and Racter even wrote a book — The Policeman’s Beard Is Half-constructed. Subtitle: “The First Book Ever Written by a Computer” which is saying something. The book is very ’80s in a specific way, lots of old Victorian lithographs collaged with computer-ish random prose.
The book itself is a bit of a put-on; there’s no way a computer “wrote it”; rather, a very clever automated form of “Mad Libs” was played and tweaked. Still, I found a copy in a used bookstore and was pretty excited. What an artifact. I went home and wrote about finding a copy on my blog (this was 2002, what can I say?). Soon after someone wrote me an email from London to ask if they could borrow it.
The book was a little rare, but it was more fun to share it. I went to the post office in Red Hook, Brooklyn, and mailed it to London. I didn’t hear a word from the person after that and figured that the book was gone and forgot about it. I mean was I going to go to the NYPD and tell them that I mailed a book written by a computer to a Londoner who’d emailed me out of the blue? But then a year or more later, the borrower mailed it back with a nice and gracious note.
I still have the book. It has a red cover. I like that someone asked, and that I sent it, and that the book has spent far more time in Europe than I have. For all the grim days and anger that we’ve seen on the Internet, the vast majority of us were, and are, over to the side quietly building a web of gentle trust. What dummies we.
This being now, you can read the whole of The Policeman’s Beard in your browser, scanned in at the Internet Archive. And you can download a copy of the software and run it in an emulated computer, as if no time had passed. It’s out there for MS-DOS, Apple, Amiga, and other platforms too.
Jorn Barger (a polyfaceted Usenet denizen, AI scholar, and coiner of the term “weblog”) maintained a Racter FAQ for many years, and it tells us that Racter was written in a custom language called INRAC. INRAC was a language for conversational interfaces and it looks like a terrible monstrosity. Reviews at the time were not favorable. I took apart the Racter disk-file and looked its contents, and found a pretty grisly set of templates with names like JOKES, RAPT, VOCAB, etc.
The mid-80s were a hot moment for chatbots. Talking to a computer was going to change the way we used computers. Conversational interfaces were on the way.
A few weeks ago I went to a dinner with many Venture Capitalists (I know that sounds terrible) and they were all quite sincerely into bots. 2016: An era of tight margins, shrinking “A” rounds, and bots. Slack, the Michael Jackson of groupware, is of course a prime host environment for parasitic bots; bots are all over Twitter, spewing automated wit, bots are answering emails and advising us. There are hilarious and inspiring bot companies. Bots are everywhere. Bots!
Chris Dixon recently wrote a well-informed article about the next waves in computing. He had this to say about “deep learning” (a set of approaches to classifying input):
Many of the papers, data sets, and software tools related to deep learning have been open sourced. This has had a democratizing effect, allowing individuals and small organizations to build powerful applications. WhatsApp was able to build a global messaging system that served 900M users with just 50 engineers, compared to the thousands of engineers that were needed for prior generations of messaging systems. This “WhatsApp effect” is now happening in AI. Software tools like Theano and TensorFlow, combined with cloud data centers for training, and inexpensive GPUs for deployment, allow small teams of engineers to build state-of-the-art AI systems.
No doubt. It is completely bananacakes out there. These advances will be the foundation for a whole wave of conversational interfaces, and a whole wave of other advances too. Lotta waves cresting.
But don’t get too ahead of things. Using Racter is not as different from using Siri as you might expect. It’s just that Siri has petabytes of stuff in her brain, whereas Racter has a floppy’s worth. Computers have changed a ton in the last 30 years, humans barely at all. Don’t mistake their progress for ours.We’ve learned how to talk to computers, and they’ve learned how to pretend to understand us. Useful when driving. People love chatting with their Amazon Echo. But the conversation still doesn’t really mean anything.
Still, bots are the next big thing. In this spirit I wrote a bot to help me write this newsletter. It just picks random elements from fun lists that I’ve pre-loaded. It’s about 50 lines of Python. I’ll open-source it once it’s further along. I’m enjoying what it does right now. We’ll give it a mini-column in the newsletter and see where things go: