FEATURE

Inventive problem solving

Ray Girvan investigates TRIZ; a methodology for formalising the invention process according to empirical rules

The former Soviet Union has left a mixed legacy to science and technology. At one extreme, it stands out for the spectacular ideological mistake of Lysenkoism (though with the growth of 'creation science', the West is no position to be too complacent about such scientific wrong turnings). At the other, its scientists earned a string of Nobel prizes in fundamental chemistry and physics: for instance, Semenov for his work on chemical kinetics, and Basov and Prokhorov for their work on masers.

Apart from this 'big science', the Soviet era has something of a reputation for producing clever oddities as spin-offs from military research, such as the Ekranoplan ground-effect craft and internal laser-etching of paperweights. It's unusual, however, to see a non-military Soviet spin-off achieving cult commercial status, but over the past decade this has happened with the invention system, TRIZ.

TRIZ is a Russian acronym for 'Teoriya Resheniya Izobreatatelskikh Zadatch' - Theory of Inventive Problem Solving - a methodology devised by Azerbaijani scientist and writer Genrich Saulovitch Altshuller (1926-1998). In contrast to techniques such as brainstorming, based on random idea generation, TRIZ aims to create an algorithmic approach to the invention and refinement of systems. Altshuller's study of engineering patents, initially when working in 1946 as a naval clerk in Baku, Azerbaijan (then in the Soviet Union), led him to believe that technological evolution is predictable. This predictability, he concluded, manifests both as nine general laws (for instance, systems evolve 'toward increasing ideality' and 'from rigid structures toward flexible ones') and 40 'inventive principles' (such as feedback, equipotentiality, and disposability) that apply across different disciplines.

  • TRIZ aimes to create an algorithmic approach to the invention and refinement of systems, rather that using techniques such as brainstorming, based on random idea generation

One of the central ideas of TRIZ is a design analysis to find areas of conflict between improving and worsening features, the outcome being a predictable subset of inventive principles that resolve that conflict. For instance, you may have a situation where bright light is a necessary function, but has a downside of fading a material. Cross-referencing 'illumination intensity' against 'stability of composition' on the TRIZ 'contradiction matrix' immediately returns ready-made resolutions: optimal property changes (i.e. shielding); non-uniform material (i.e. local shielding); and disposability (i.e. let the fading happen, but replace the component). Altshuller also developed ARIZ, a systematic algorithm for working through an innovation project using TRIZ principles. Patent analysis was another important area; for a given invention, he identified a characteristic 'S-curve' of numbers of patents versus time, corresponding to the development of that invention along the evolutionary path toward ideality.

It took half a century for TRIZ to reach its current vogue. Altshuller began promoting his ideas in the 1950s, on his release after four years in the Vorkuta Gulag for questioning Stalin's patent policies. Living on income from pensions and writing science fiction, he spread his views on invention through seminars, papers and a series of books, helped by a core group of Trizniks largely drawn from Soviet Jewish scientists. Despite bureaucratic inertia, interest in TRIZ grew steadily within the Soviet Union until the perestroika era, when the state committees for invention finally endorsed TRIZ. It also reached the West via Altshuller's ex-students, other enthusiasts, and book translations. This period saw the first TRIZ software, as well as the diversification of 'classical' TRIZ into competing and often radically changed forms. For instance, one of the major consultancies, James Kowalick's Present Vision, has expanded and rebadged TRIZ as 'Next Generation Thinking'. Now, five years after Altshuller's death, many international companies are selling TRIZ consultancy and products to the corporate technical market.

  • TRIZ-based packages are very similar, with an input stage allowing the user to specify the system to be invented or improved, followed by a knowledge-base stage that applies known solutions and data to the problem

Given its algorithmic nature, it was a fairly inevitable development that TRIZ could be implemented in software. Generally, TRIZ-based packages (all of the following being for Microsoft Windows) are similar in structure. An input stage lets you specify the system to be invented or improved, followed by a knowledge-base stage that applies known TRIZ solutions and external data to problems arising. TRIZ Explorer 1.5, from the Dutch firm Insytec, is an exception in providing the knowledge base alone. Running inside Microsoft Internet Explorer, it's a structured TRIZ reference-tree, covering inventive principles, inventive standards (i.e. generic patterns of inventive solutions), and physical effects. The 'leaves' of the tree are explanations and examples, which can be expanded with users' own MS Office files or links to Web pages.

TriSolver Professional, the 'Idea Generator and Manager' originally developed as an in-house tool for the TriSolver Group, Hannover, is more typical. Its function is to act as an 'idea pool' for managing material, perhaps generated during a brainstorming workshop, in multiple document formats such as PDF, Word, Excel, and HTML. As with TRIZ Explorer, this is coupled with a categorised tree of TRIZ concepts, but with further tools such as the TRIZ Contradiction Matrix and TriSolver's adaptation of ARIZ. TriSolver's customer base is largely within Germany, but it comes in German and English versions.

The Belgian company Creax offers a TRIZ package within a general ethos of creativity for innovation. Its Innovation Suite comprises three modules: CreaTRIZ Classic, for technical applications; CreaTRIZ, for business and management; and Evolutionary Potential. Recommended by the TRIZ Journal as 'most suitable for pedagogic use', it provides a self-explanatory sequential format for TRIZ invention. CreaTRIZ first guides you through defining the desired system, modelling it as a functional block diagram, then analysing how to nudge it toward ideality by drawing on standard TRIZ tools - contradiction resolution, trends, and so on - and online sources including European, US, Canadian and Japanese patent gateways. One source, the Function Database of physical effects, can be tried online.

In the USA, two of the most significant developers of TRIZ software are Ideation International and Invention Machine. The first has strong roots in TRIZ's Soviet past - its founders including Boris Zlotin and Alla Zusman, who ran a TRIZ technical school in Kishinev, Moldova. Economic conditions led to their emigration to the USA, where they co-founded the Michigan-based Ideation in 1992, bringing many of their Kishinev colleagues to the team. Their Innovation Workbench is based on I-TRIZ: TRIZ updated with integrated tools, problem formulation modules, a larger knowledge base, and new applications such as Anticipatory Failure Determination (failure analysis and prediction). Like CreaTRIZ, it first helps define the problem, through analytical tools such as questionnaires and ARIZ, then applies knowledge-based tools to synthesise a solution. IWB covers applications throughout the design cycle: 'design, research and development, manufacturing, safety, reliability, and quality assurance'.

Innovation Machine, based in Boston, has a narrower brief, specialising in supporting the pre-CAD stage of technical development. In addition, however, its Goldfire service has extremely sophisticated data-mining tools for patent analysis, extending its scope to general technical searches and the management of scientific intellectual property. I was recently shown a preview of its forthcoming Goldfire Innovator; see the panel for more details.

I admit to a certain caution about TRIZ. How inevitable is its course of system evolution? MEMS micromachines, for example, by replacing electronic components go against the evolutionary trend of mechanical to electronic. Is it an empirically justified methodology, or an unproven ideology? There are strong similarities, for instance, to Marxist dialectics. Both Marxism and TRIZ hold that systems evolve toward perfection and that history repeats itself, and Marxism's dictum 'thesis - antithesis - synthesis' resembles the TRIZ resolution of contradictions.

The TRIZ market has its dissidents, notably Yevgeny Karasik, one of Altshuller's ex-students and a classical Triznik, who is highly critical of current offshoots from the original vision of a general, non-computerised, approach to innovation. He notes that 'There is still no proof that algorithmisation of problem solving - an art - is achievable. Altshuller's research technique is hard to reproduce effectively, as nobody is as good at formalising the intuitive process of extracting interesting and novel inventions from the ocean of all inventions. However, that aim is appealing; there's an objective necessity to significantly improve the innovation process in industry. TRIZ (and all its modifications) promises to fill that vacuum; that is why it finds clients.'

Though Karasik is in a minority, I think he acts as a very necessary gadfly in ridiculing the excessive claims for TRIZ, such as the fallacious argument of retrofitting it to innovations found in other ways ('TRIZ would have done it this way, therefore TRIZ works'). Another problem is finding case studies. Despite the many major companies cited as using it, it's hard to find concrete examples akin to those, say, publicising particular applications of mathematics packages. TRIZ suppliers argue that this is largely down to issues of company secrecy over intellectual property. Brian Campbell, Innovation Engineer for Creax, told me: 'We do need more case studies to improve the credibility of TRIZ. But some people are coming forward, such as Ian Mitchell of Ilford Imaging - a small company, showing it's not just for corporates - who documented how he used TechOptimizer to improve a photo emulsion coating process.'

'The key drawback, to my mind, is the name,' Campbell added. 'The fact that it's a Russian acronym doesn't help; the literal translation is inaccurate. While it's inventive, it's not merely a theory; it has solid empirical backing. And it can be applied to more than problem solving. Current trends are to apply it to wider issues such as social and business, and to incorporate it into Six Sigma [the quality management system used successfully by Motorola and other US companies]. On the UK academic circuit, Darell Mann [Creax director and president of the European TRIZ Association] is using it at the University of Bath, and the Plymouth University BSc/BEng course is starting to teach TRIZ alongside other systems, such as brainstorming and Quality Function Deployment, for solving design problems. Our inaugural meeting of TRIZ South-West, a regional group to discuss TRIZ, was there on December 3rd.'

This still leaves me with doubts about the fringes of TRIZ, such as its application to politics, or claims that it could have predicted the 9/11 World Trade Centre attack. However, its strengths in its core area of scientific and technical innovation look very tangible.

For more information about TRIZ, see the Altshuller Institute for TRIZ Studies, 'the only TRIZ organisation in the West officially authorised by Genrich Altshuller', and the TRIZ Journal.

Invention Machine offers two packages

 

Invention Machine, with over 1,000 installations worldwide and customers including Boeing, Honda, HP, Procter & Gamble, and Samsung, describes itself as the premier provider of Enterprise Innovation Process software and services to the Global 2000. It has been some five years since I last looked at its TRIZ-based invention packages - at the time, called TechOptimizer 3.0 and IM-Phenomenon 1.0 - but the implementation has changed dramatically. At the time of writing, Invention Machine has two packages: TechOptimizer software for TRIZ-style invention; and Goldfire Intelligence, an online patent and knowledge database service. This, however, is changing with an integrated release scheduled for early 2004, as Goldfire Innovator.

Goldfire is a search engine accessed through a Web browser. It draws on data sources including multiple patent servers; a quoted 2,000 technical websites, many on the Deep Web; Invention Machine's own database of 8,000 scientific effects; and your own knowledge base if you've created one. A search can be targeted at everything online or a subset, and you can append anything you find to a report for later study. The non-standard aspect, that became obvious when I tried a test drive, is that this isn't an ordinary search; Innovation Machine has semantically analysed and parsed the texts to identify 'subject, action, object' constructions, optimising the system for finding answers to natural language technical questions. 'How to make a humane mousetrap?' 'What oxidises gold?' There are patents telling you how. More impressively, any of the texts recovered have been 'distilled' for reading at five user-selectable levels of detail, making it easy to skim then zoom in for closer study.

Neil Sherry of Invention Machine explained: 'This is the biggest use by volume. Around 80% of information on how to do things in the technical and scientific field is obtainable from patents.'

Further options come under Innovation Trend Analysis: company profiles in terms of patent activity, comparison of companies, technology trends by patent class, and forward and backward chains of citation. This serves users whose interest is in intellectual property and market intelligence; for example, deciding if a concept is worth exploring or has been mined out, or to make 'patent hedges' to protect your company's future plans or hinder competitors. (Patent trends link in with TRIZ, since its laws of technological evolution predict where patents may go; if, say, a manual device exists, an electronic version may well follow soon).

Goldfire Innovator will closely integrate TechOptimizer (as Optimizer) and Goldfire Intelligence (as Researcher). I saw a brief demonstration showing how they work together. Optimizer, kept on your local machine, is somewhat similar to Creax up to the stage of your creating a functional block diagram. However, the automation and polish is much greater. If you decide to trim an unnecessary component (a standard TRIZ process for improving a design) Optimizer automatically relinks to adjacent components and displays any contradiction that needs to be resolved. The elegant part is that when, as intended, you work while online, Researcher automatically searches in the background for knowledge that provides this resolution. Optimizer will also contain new functions such as Root Cause Analysis and a Patterns module to facilitate modifying a system to its next evolutionary stage as predicted by TRIZ.

Innovation Machine doesn't stress its TRIZ origins. Neil said: 'It's partly to avoid TRIZ-geekery, partly because it's not merely based on TRIZ. For instance, the process for removing unnecessary components, although it's similar to the TRIZ 'trimming', comes as much from Value-Engineering Analysis.'

Ultimately, Invention Machine's main marketing point is that its software increases the conversion rate of ideas-to-products. According to a recent Economist article, 'Innovation by numbers' (19 June 2003) it typically takes 3,000 ideas to develop 100 worthwhile projects, which may well lead to only one finished product. Anything that improves that track record will be financially beneficial for companies. However, the purely scientific benefits of natural language information retrieval and the creation of local knowledge bases shouldn't be underestimated.

Feature

Robert Roe reports on developments in AI that are helping to shape the future of high performance computing technology at the International Supercomputing Conference

Feature

James Reinders is a parallel programming and HPC expert with more than 27 years’ experience working for Intel until his retirement in 2017. In this article Reinders gives his take on the use of roofline estimation as a tool for code optimisation in HPC

Feature

Sophia Ktori concludes her two-part series exploring the use of laboratory informatics software in regulated industries.

Feature

As storage technology adapts to changing HPC workloads, Robert Roe looks at the technologies that could help to enhance performance and accessibility of
storage in HPC

Feature

By using simulation software, road bike manufacturers can deliver higher performance products in less time and at a lower cost than previously achievable, as Keely Portway discovers