CTO Blog

CTO Blog

Opinions expressed on this blog reflect the writer’s views and not the position of the Capgemini Group

Rocket Science

Is there always something new to wish for? In some cases, you are inclined to think not. Take for example Charles Simonyi, an above average gifted programmer from Hungary who joined a modest start-up company in the beginning of the eighties that would later headquarter in Redmond. But not before working at Xerox Parc with IT legends legends such as Alan Kay and Robert Metcalfe on some, well, interesting projects (the first WYSIWIG text editor in the world, somebody had to come up with it). If you joined Microsoft before the PC existed and you sort of single-handedly created both Word and Excel, you may claim some street credibility in the IT profession. Financially, you are probably doing quite well too: Simonyi is an active philanthropist, with an emphasis on supporting art and culture. Clearly he saved a little for himself too, as he made headlines in the past two weeks as the tourist who paid 18 million Euros to fly with a Soyuz to the International Space Station. The crew had plenty of work to do there, so no doubt Simonyi had oceans of free time during his stay. There must have been some contemplative moments, Charles having a look through the porthole at Mother Earth, circling underneath. Reached the absolute top in the profession. In addition, made a virtually impossible boys dream come true. And he might still be asking himself if there is a new daring challenge somewhere, waiting to be cracked. Fortunately, it looks like Simonyi already has all the clues on where to search. In 2002, he left Microsoft to pursue with his own company the answer to the question that has been haunting the IT industry ever since the very first beginning: why is it so terribly difficult to develop good software? Rapid successions of programming languages, visual development environments, case tools, formal specifications, object-orientation, frameworks, agile methodologies, model-driven architecture: whatever we come up with, it seems that the gap between dream and reality - between intention and solution – always stays the same. I just celebrated my tenth anniversary as a jury member of a prominent programmers contest in the Benelux (the ‘RAD race’) and I couldn’t escape the feeling that the productivity of the teams more or less stayed the same during the course of the years. Of course, systems are more complex than ever and we expect better user interfaces and superior services. But it seems as hard as ever – at best – to meet the expectations. It is still painfully difficult to describe business requirements and translate them into working solutions that truly mirror the initial ideas. Charles Simonyi thinks he has found the answer in an approach that he has coined Intentional Software: provide all the support that is needed to capture the real intention of a solution. Use language concepts – visual, textual – that were meant for that from the ground up. Jargon from the pension world or logistics, actuarial formulas, tables, sketches, drawings: whatever makes a domain expert feel comfortable is suitable. Record the intention in a format that is perfectly independent from its rendering (Simonyi fancies meta concepts) and gape in admiration at all the different representations that allow you to view and edit the intention, all depending on the audience you are targeting for communication. One of the representations – almost a detail – is code: comes in particularly handy when executing on a computer. And it truly exists too. We are currently piloting an early version of Simonyi's language workbench and we find that both hardcore software engineers and business domain experts are equally enthusiastic about the approach. Which is not exactly a déjà vu. You might call it the holy grail of the IT profession: communicating with end users and domain experts in their own language and seamlessly generating systems from that. Without any distortion in between. We have hoped for it before, for example when COBOL and 4GL’s arrived (“from now on, managers can code themselves”). Sometimes, we are very far away: UML 2 is even too complex for its own creators. But nothing stops us from dreaming. In practice, becoming an astronaut does not even involve rocket science. Who knows, one of these days we might even find the grail.

About the author

Ron Tolido

Leave a comment

Your email address will not be published. Required fields are marked *.