Wolfram bros seek code slinger posse for IBM and Google round-up
Come in Google, your time is up. Watson, we're watching you
The brothers behind the powerful Wolfram Alpha search platform and Wolfram programming language want code-slingers to take on IBM and Google.
Stephen and Conrad Wolfram, US and European CEOs respectively of Wolfram Research, are trying to attract Java, Python and Ruby programmers to Wolfram-as-a-cloud.
The goal is for those versed in traditional languages to adapt to its algorithmically rich Wolfram Language and associated query engine, Alpha.
Strategic director of Wolfram Research US and CEO of Research Europe, Conrad told The Reg: “We will be working on programs so it’s easy to transition.”
Wolfram Language is a combination of the Alpha search engine and Mathematica, also a Wolfram Research language created by Stephen who also founded Wolfram Research. Alpha is the powerful natural-language search site released by British computer scientist and entrepreneur Stephen in 2009.
Alpha differs from Google, which indexes sources and finds answers to already asked questions or pulls back known sources of information as possible answers to a query in a long list. Instead, Alpha tries to answer new questions and to then present that data in a variety of different ways at the same time – via text, charts and visuals.
So why should devs happily versed in Java, Python or Ruby want to bother with yet another language – and one owned by such a highly specialised firm?
Conrad reckons if you’re doing anything that demands intense computation then Wolfram Language will let you program between five to 10 times faster.
“We keep finding the financial people who start moving from Java to us say they can do cluster analysis really easily, so they start building in these much higher things. There’s some learning curve and one of the problem is getting people to recognise there’s more to learn to get to that higher speed,” he told The Register.
Conrad, younger brother of Stephen, founded Wolfram Research Europe in 1991. Yet Wolfram Research’s push is not entirely philanthropic: the firm is making a play to broaden the use of its 25-year-old language.
The vehicle for this is the cloud – and Wolfram has been offering hosted versions of its Alpha engine behind customers' firewalls plus a Wolfram Language cloud. In addition, Wolfram has been letting customers run private instances of Alpha, installing its engine and then importing their own to create a private version of Alpha.
It’s a model the Wolfram brothers now seem to be pitching as a kind of decision support and advanced analytics engine for businesses. Conrad said customers span financial, pharmaceutical, health, media, marketing, telecoms and government.
Yet Wolfram isn’t alone in this: Metanautix, for example, is a VC-funded start-up founded by an ex-Google and Facebook duo, is building Quest – which they call a “data compute engine”.
The idea is Quest hovers above a range of structured, unstructured, relational and other data sources; you query Quest via a Tableau UI and it pulls back the results. Quest is written in C++ married to a Java Virtual Machine which potentially lets you plug in JVM-friendly languages, such as Java and Python.
Quest works with MongoDB, with plans to integrate with SAP and Salesforce.
“We’ve got the slice through computation,” Conrad reckoned. “I think of what we have is like a computation layer. In ancient times there were ERP players... I think we are in the realm where you need a layer to do enterprise computation.”
There’s also IBM. Conrad claimed Wolfram and Alpha have been seeing off IBM’s Watson in customer deals that involved search and machine learning. Watson is IBM’s supposed artificial-intelligence system able to answer natural-language queries and that IBM’s decided to turn into a paying product.
“They have a different process… they have a machine learning process, Google has a search process. We have a computational process. I think their [IBM] process often is troublesome to make it work in the situations where they’d like it to work.”
The Wolfram platform doesn’t come lightly. Google indexes existing sources while Wolfram Research conducts what is essentially a massive ETL exercise: importing public data to Alpha and cleansing it to make it – in Conrad’s words – “computable.” This is what happens on the public Alpha site and customers’ private sites.
The Wolfram Language works using a set of 5,000 functions and, what the company claims, is one of the world’s largest collection of algorithms – “tens of thousands” plus “hundreds of millions” of pieces of information.
The language spans core language and structure, data manipulation and analysis, symbolic and numeric computation, time-related computation, computation of geographic data and the production of graphs, networks images and sound.
Wolfram claims not to separate computation from data or its representation – unlike these other languages. The Wolfram language is executed through Wolfram’s Computable Document Format (CDF) Player, a run-time library. Data, formulas, code, graphics and everything else are represented as expressions.
In the last year, Wolfram has rolled out a data-science platform and a publishing program, Mathematica on Raspberry Pi, while it’s also released a programming cloud letting you use Wolfram Language on desktop and mobile. This week, Stephen Wolfram was pitching apps written using the Language on Apple’s Watch. The Wolfram clouds run on Amazon's AWS.
Wolfram Language is proprietary a fact the firm reckons it can get around by having the language and Alpha run on other people’s clouds – bringing it to them.
Also, the Wolfram SDK plugs into the Eclipse workbench, with hopes for a big Wolfram ecosystem of plug-in providers. The Wolfram platform runs on Windows and Linux, Intel and ARM and work with frameworks such as Hadoop.
Conrad believes the pure search, as epitomized by Google, has gone about as far as it can and that the future is now belongs to computational search – hence the push and new-developer sign up.
“Search has had a huge amount of money behind it and it does very well but I’m not sure how much further you can go with pure search.
"Anything that has computational ways to find answers goes a long way to finding what companies want. How far on the continuum of what we could achieve have we come? Computation, 10 per cent, [Google] search 70 or 80 per cent.
“There’s more room in computation to get things done.” ®