Talking to Computers

Talking to Computers

Adapted from an article by

physicist and programmer Stephen Wolfram 

published in Eureka, 2013

http://www.mathigon.org/eureka.pdf

Do we really need computer languages

to tell computers what to do?

Why can't we just use 

natural human languages,

like English or Spanish, instead?

If one would have asked this question

a few years ago, most of the experts

would have said it was hopeless.

That perhaps one could make toy examples,

but that ultimately natural language

just wouldn't be up to the task

of creating useful programs.

But then came along programs 

like Wolfram | Alpha in which

they have been able to make

free-form linguistics work, vastly better, 

than one ever thought possible.

But still, the input is essentially

just set up to request knowledge

- and responds 

by computing and presenting

whatever knowledge is requested.

But programming is different.

It is not about 

generating static knowledge,

but about generating programs

that can take a range of inputs,

and dynamically perform operations.

The first question is:

how might we represent these programs?

In principle one 

could use pretty much

any programming language.

But to make things practical,

particularly at the beginning,

we need a programming language

with a couple of characteristics.

The most important is that programs

a user might specify with short pieces

of natural language must typically 

be short - and readable - 

in the computer language.

Because otherwise the user

won't be able to tell

- at least not easily -

whether the program

that's been produced

actually does what they want.

A second, 

somewhat related criterion

is that it must be possible 

for arbitrary program fragments

to stand alone - so that

large programs can realistically 

be built up incrementally,

much like a description 

in natural language

is built up incrementally

with sentences and the like.

To get the first of these characteristics

requires a very high-level language,

in which there are already many constructs

already built in to the language -

and well enough designed

that they all fit together

without messy "glue" code.

And to get the second characteristic

essentially requires a symbolic language,

in which any piece of any program

is always a meaningful symbolic expression.

Conveniently enough, 

there is at least one language

that satisfies rather well 

both these requirements: Mathematica.

The linguistic capabilities of Wolfram | Alpha 

gives one the idea that one might be able to understand

free-form natural language specifications of programs.

Mathematica, for instance, 

is what gives one the idea 

that there might be a reasonable target

for programs generated automatically 

from natural language.

Stephen Wolfram express that for him,

there was also a third motivating idea

- that came from his work displayed

in his book A New Kind of Science.

One might have thought, he says,

that to perform any kind of complex task

would always require a complex program.

But what he learned there

is that simple programs 

can often do highly complex things.

And the result of this

is that is often possible

to find useful programs

just by searching for them

in the computational universe

of possible programs -a technique 

that he says he use with increasing frequency

in the actual development of both

Wolfram | Alpha and Mathematica.

__________

I transcribed Stephen Wolfram's short article

because, although his work is meritorious

he invest too much time in selling himself

and his products, every time he has the opportunity.

Probably this is the part of the way to stay alive after

thirty years in the business. It seems not enough to

work on the frontier, you have to invest continually

time, words and effort, to present

your contributions, as the last word of everything.

It seems this work for the market

but it can be a little bit tedious

when one expect ideas instead 

of the speech of a salesman.