When Wolfram Research launched its “computational knowledge engine” called Wolfram Alpha way back in May 2009 to much fanfare, it sounded as if time-travel had arrived, but it also confused a lot of people who compared Alpha (a little unfavourably) with search engines such as Google and Yahoo.

Some industry commentators, however, perceived that it was rather like saying that a screwdriver is a bad hammer – Alpha, they said, is not a generalised search engine in the mould of a web-trawler such as Google.

 Instead, what Alpha does is curate data sets with financial, mathematical, scientific and other data that you can query it using simple questions or that you can manipulate using mathematical formulae.  These data sets are used with Mathematica, the flagship of Wolfram Research and well-known in academic circles where it is used to perform complex calculations, manipulate data and create graphs and visualisations. 

Mathematica was once described as the ‘ultimate ‘discovery interface allowing people to sloppily mix all sorts of notations with fragments of natural language.’

Conrad Wolfram founded Wolfram Research in 1991. It is here that  he initiates and directs business, marketing, design and a variety of other strategic technical projects-everything from the Mathematica Player family concept to webMathematica (to conceptualizing wolfram.com).

‘I am interested in how technology and computation can move our lives and economies forward and am campaigning for the reform of mathematics education based on computers.’ 

He is also claims to be the world’s leading advocate for a fundamental shift of maths education, by founding computerbasedmath.org.

Conrad is by no means an obsessive mathematician thought he admits to an MA from Cambridge University where he read physics and mathematic:  He’s also  an avid photographer, he plays the piano pretty well and  he’s been a featured speaker at TED.

RM:

Conrad, would you take me back to how Wolfram|Alpha was conceived. Did the site cost a vast amount of money to set-up? How did you program it?

CW:

Wolfram|Alpha is built on 20+ years of our technology, much of it cutting edge. Since 1988 we have been the leading “computation company”, pushing the envelope of what the paradigm of computation makes possible. We realised that we could apply this to knowledge and that there would be many benefits of “computable knowledge”–the paradigm of Wolfram|Alpha.

Wolfram|Alpha is mostly programmed using Mathematica, and deployed using grids of webMathematicas, so it’s very much based on the 20+ years of R&D. So there’s no question that it is a large and expensive project but that it’s been possible at all is only because Mathematica technology has increasingly integrated advanced pattern-matching language with raw computational power, automation at every level and now computable documents.

RM:

When you launched did you have a positive response, pretty quickly?

CW:

Yes, people saw we’d done something different, something significant. They saw there was a different way to interact with knowledge that wasn’t search, though plenty confused it with this.

RM:

How much human input is there in Wolfram Alpha?

CW:

It’s a complicated process. When we suck the knowledge in, it’s a mixture. We’re getting better at automating this process, but there are humans who actually look at it and try to figure out how to structure the information and curate it and make it accurate.

There’s a fair amount of human input, but there’s increasing automation in that sort of scanning process.

This is a general theme I guess we have. Part of Wolfram Alpha is encoding expertise of humans. We don’t feel we can do it all by being smarter or having the right algorithms. We can absolutely assist that, but it’s automated assistance of humans, rather than necessarily doing the whole thing, in terms of getting the information in, automatically or by computer.

Humans are consuming this information at the very end, so one wants to automate the process very much in between. One also wants to figure out what it is humans are interested in knowing, at some level.

RM:

The starting point of Wolfram|Alpha is that people increasingly want the internet to provide them with correct answers to questions, rather than, providing a selection of pages you have to read before deciding whether or not they are relevant.

Is there a problem here that fewer people will bother to learn or research certain material and rely instead on the internet?

CW:

I really don’t think that’s how it will play out. It’s true that machinery and automation mean we may not always learn the facts or manual procedures that earlier generations have, but in exchange we can operate at a much higher level, covering much greater ground. Inquisitiveness and research is at a higher level–and often at a higher rate. What’s critical is that people gain experience to apply appropriate questioning and understanding to the results they get.  

RM:

Can you remember anything that surprised you about the early days of the site?

CW:

How revolutionary it seemed to people for math education, and its sudden effect in moving the agenda of computer-based math education forward (and a help to my starting of www.computerbasedmath.org).

 For years Mathematica has offered great math power for education, so doing explicit math was the aspect we thought least revolutionary about Wolfram|Alpha. But I think the openness of rough-and-ready input that it was freely available on the web made for this qualitative change in perception if not yet of practice of a ubiquity of integrating computers into math education.

Siri’s use of Wolfram|Alpha for knowledge (and math) queries takes this ubiquity of interface further still.

RM:

How have you grown the database of searches? Have you added features based on user feedback?

CW:

Yes, we aggregate and depersonalise info on queries and this informs improvements for linguistic recognition, domains to cover, data improvements in those areas and their presentation, which itself is an important task.

RM:

What is the most intellectually challenging issue for Wolfram|Alpha?

CW:

”
… perhaps the nub of the challenge is
how to represent knowledge in a way
that’s computable, not just semantic

There are quite a few, but perhaps the nub of the challenge is how to represent knowledge in a way that’s computable, not just semantic. We use the very versatile symbolic expressions for literally everything from computations to documents to interactivity to data.

RM:

Is there anything about Wolfram|Alpha that users have misunderstood?

CW:

People do refer to it as a search engine, for example something which matches other people’s words on the web. It’s not doing that. Instead when you enter a query, we interpret that query then compute a result from pre-curated knowledge. We’re generating new custom answers to your questions, not searching nearest matches of other people’s information. So knowledge engine is a better descriptor.

RM:

Do you think that technology has much to learn from other branches of Science, which have made occasional ‘breakthrough’ progress as a result of externally or internally promoted Grand Challenges? Is Wolfram Alpha tackling one of these, how to make computers understand natural language?

Worldwide collaboration is now the norm in physics and astronomy; and more recently in genetics and molecular biology. It was the international Grand Challenge of the Human Genome project that triggered the necessary shift in the research ethos of biologists towards the larger scale and the longer term accumulation of knowledge. After twenty years, its results are beginning to trickle through to medical practice.

Is not a similar breakthrough now in prospect for the verification of software?

CW:

Clean design is crucial to the usability of technology. Apple has made this very obvious. And good technology design is rarely by broad collaboration, usually—at its nub—a very small group, even an individual. Projects like the Human Genome have a very specific output. Competitions can work well in cases like that. When you want general or open-ended achievements—like make knowledge accessible, you need clear vision for getting to that. Once sufficiently established, collaboration can work well.  Perhaps there is room for a Grand Challenge in this area, but if so, we haven’t yet conceptualised it.

RM:

You’re a great believer in that you can teach children to be better at maths by teaching it through computer programming. There’s also an argument which starts that programming teaches a way of thinking that’s important, it gives you order and structure in life. Opponents, say programmers are completely misunderstanding the world in exactly the same way everybody else does and it doesn’t make you intellectually superior. What camp are you in?

CW:

”
To be clear, there
 are many facets to
intellectual understanding.

I’m all for programming. It’s a crucial way to express ideas–a vital skill. To be clear, there are many facets to intellectual understanding. Learning to program isn’t the only facet of math,  but it’s an important one: it’s the richest way to express concepts. Another aspect of learning math comes from the other extreme to building programs up from scratch: using other people’s complex black box models with appropriate critical understanding.

That’s a crucial skill too. Right now most math education steers a middle course of neither being able to build up sufficiently complex problems from scratch nor being experienced to be able to work with complex models that others have made. Applied correctly, computers and programming can dramatically uptick the level of both and most importantly add conceptual and as well practical empowerment.

RM:

I was reading an interview with your brother Stephen where he is quoted as saying: “The thing I tend to do is take these big complicated things and try to drill down to get the fundamental components underneath,” Do you think that technology, and our very civilization, will get more and more  complex until it collapses? There’s no opposing pressure to limit this growth.

CW:

I don’t see why complexity should lead to collapse. But with each new tranche of complexity come benefits and challenges. Screw-ups occur (sometimes very bad ones–like the financial crisis) which sooner or later get adjusted for – at least that’s the world’s history in aggregate so far.

RM:

Do you think it’s right that people judge the value of technology more by its economic effects than by its intellectual effects and computer scientists by applications rather than by contributions to knowledge, even though contributions to knowledge are the necessary ingredient to make previously unthinkable applications possible?

CW:

People should judge achievements as they see fit, but seeing, utilising, designing for and applying the growing multiplicity of technologies is a major intellectual challenge, perhaps the major intellectual challenge of our age. And what makes it particularly hard is the open-endedness of what’s now achievable. Technologies criss-cross, ‘anything’s possible – the problem is to figure out what you want to do, which most of today’s education does not prepare us for well. Open-ended problems are usually the hardest of all.