Can Nothing Stop Computo!

by John Holbo on July 25, 2007

I present you with this week’s reason for thinking I read too many Legion of Super Heroes comics. I’m reading Crangle and Suppes, Language and Learning for Robots. (It’s part of the super popular ‘for robots’ series. My favorite being Wine Tasting For Robots.)


This book reports research that the authors have been doing together for the past decade on instructible robots …

In this book we explore the following two specific questions. What does it take for a robot to understand instructions expressed in a natural language such as English? What further challenges arise when the robot must learn from that instruction? The work we present falls naturally into three parts: theory, language performance, and learning. We briefly summarize each of these parts.

Part I on theory consists of four chapters. The first one sets forth our general ideas about instructible robots and how they are different from robots that operate autonomously. (xv)

Yes, but why do they have to be indestructible, I was asking myself?

I actually continued on like this for some time. You see, I had me a brief little thought about how it was probably going to turn out that they had some very abstract story about what a robot could learn, given an infinite amount of time. And somewhere along the line someone decided ‘indestructible robot’ was cute shorthand for this ideal limit; and I just never got the memo. Anyway. I think I need to get some sleep.

Likely culprit.



quercus 07.25.07 at 2:03 pm

Hmm. I actually read it as ‘indestructible’ too.
But my mind was otherwise occupied with wondering how “theory, language, language performance, and learning” add up to three parts (my all-too-destructible mind keeps getting four parts).


Kieran Healy 07.25.07 at 2:10 pm

People claiming there are no robots may themselves be robots.


John Holbo 07.25.07 at 2:24 pm

I had typed the passage wrong, quercus. Corrected now.


derek 07.25.07 at 2:54 pm

There is a web site called, that specialises in reader-supplied how-tos in step-by-step form. I keep waiting for the Instructable Captain Scarlet, but no luck.


ajay 07.25.07 at 3:22 pm

From the headline, I was expecting something to do with Alan Turing and the halting problem…

— Can nothing stop Computo?
— I don’t know. In fact, it’s logically impossible for me to know, in general terms, whether Computo will stop or not.


Hermenauta 07.25.07 at 3:28 pm

Subliminal advertising from “Transformers”?

#5: :)


John Holbo 07.25.07 at 3:44 pm

Actually, as to the ‘can nothing stop’ problem, I was chuckling quietly to myself as I often do – they said I was mad! MAD! to write this post! – because chapter 7 is “Saying ‘stop’ to a robot”.


dr 07.25.07 at 4:37 pm

I also read it as ‘indestructable’, assuming that ‘indestructable’ was a term of art describing some formal aspect of robot brain pans.


magistra 07.26.07 at 6:42 am

I read it as ‘indestructable’ too at first, which I think proves that it is a badly worded sentence. Partly it’s that ‘instructible’ is an uncommon word – I had to look it up in my dictionary to check it was a standard word and not just an invention of the authors. But it’s also that ‘indestructable robot’ makes sense as a phrase, while ‘instructible robot’ doesn’t. In a fuzzy sense I think of all robots as being instructible, because you tell them what to do, so that the phrase come across as tautological. It’s only when you add the provisio of the use of natural language, that you can see what the distinctiveness of these robots is.


asm 07.26.07 at 8:44 pm

Somehow, I did not have this particular problem. The opening sentences introducing “instructions” in a “natural language” suggested to me the sense of the word “instructible” in terms of “natural language instructions”, disambiguating them from “regular” robots who read some type of machine instruction/code or the other.

I think it has to do with what words and phrases parse as “technical terms”, and not merely English words and phrases, in your mind as you read the passage. (I remember reading a small bit about some research on this aspect.) If you mentally parse “instruction” as a technical term, then the word “instructible” might then parse as an adjectival form of this term.


Alan Bostick 07.26.07 at 9:00 pm

Domo arigato, Mister Computo.


Joshua Glenn 07.27.07 at 3:04 pm

This is very funny, John! And just yesterday I was holding volume 1 of DC’s bound reissue of the LSH comics. If the pages had been in color, I might have bought it. I was once a big, big fan.

Comments on this entry are closed.