TRIZ – Theory of Inventive Problem Solving | Hacker News Reader

TRIZ – Theory of Inventive Problem Solving

en.wikipedia.org

58 pointsd99kris16 comments

Thank you for sharing this OP, never heard about it before.

Christopher Alexander (architect and influencer of GoF's original software Design Patterns) very cleverly re-frames design problems in one of his books [1] as optimization problems, where the designer is tasked with minimizing the tension between two or more conflicting requirements arranged as a graph, and where each requirement interacts positively or negatively with it's neighbours. By representing a design problem this way, it follows that "solving" it amounts to clustering requirements together to form sub-components that interact well together, or redefining the solution/context boundaries to re-frame the problem and minimize the overall tension.

It's interesting how this TRIZ framework arrives at many similar insights about the design problem, and I wonder what you get by mixing the two approaches. A lot of what makes it hard to apply is also what makes Christopher's framework hard to apply, which is clear definition of the requirements in the first place. C-K theory seems to overcome these limitations but I haven't played with it yet.

[1] https://en.wikipedia.org/wiki/Notes_on_the_Synthesis_of_Form

[2] https://en.wikipedia.org/wiki/C-K_theory

Notes on the synthesis of forms is an awesome book! It really is about systems design in general - the graph based view is the concrete formulation of the fact that elements of the system affect each other amd if these connections are not made explicit in the design process then the components will still be connected - albeit in hidden and accidental ways which will make the system development into a neverending game of whac-a-mole problem fixing.

In retrospect this may sound obvious and trivial but taking this simple graph based view of the basis of systems design as the fundamental idea gives concreteness to e.g. lot of software systems practices and unifies them.

Fascinating reading, thanks for that.

ah damn, I've already bought two Christopher Alexander books ("The Timeless Way of Building"[1] and "A Pattern Language"), and now I'm going to have to get another one. That sounds great.

[1] - well worth a read. Did you know there was a terribly humane philosophy behind the original design patterns work? This was beautifully described in Peopleware as "Local control of design by those who will occupy the space". E.g. - Alexander advocates for the people who are going to live in and work in the buildings be intimately involved in both the design and the construction of the buildings. Not only does he advocate for this, he explains ways of doing this in practice. This philosophy could be ported to software -- let the users of software systems be involved in the design and construction of the systems. That sounds like far more of an ambitious and genuinely positive idea than the old "visitor pattern" / "singleton [anti]pattern" stuff that springs to mind when one has been subjected to the computer-science-education version of design patterns.

Recently a book for kids was published, in Russian, which describes TRIZ in an easy manner.

http://fictionbook.ru/static/trials/09/81/13/09811350.html

Main character, a boy called Denis frequently sees problems around him, which allow him to apply his TRIZ skills. The book has explanation and problems, so the reader can try to learn the method.

404 Not Found

Works for me :( . Here are other links -

http://5razvorotov.livejournal.com/1197775.html http://www.u-mama.ru/read/section/reviews/books-for-child/72... http://www.labirint.ru/books/376992/

Man, I really want to read this. Wonder if there's an English version. Is it worth reading?

I'd say yes it's worth reading. The material is pretty basic, as it's targeted on 10 years old. The book came I think only in 2012, I'm not sure if there is an English version.

The Wikipedia article has no criticism section.

It might be great as taxonomy of problem solutions, but there is no reliable evidence that it works as a methodology. Reducing research activity to a bunch of codified chess-like moves is a huge temptation. However I haven't seen its adepts producing any unexpected results (in practice, not on some TRIZ textbook puzzle). The solutions were on par with what you'd expect from a person of comparable intelligence, experience and education background otherwise.

I learned it as part of my 2nd year mechanical engineering design course. It was a great way to teach concept generation to novices and expand awareness of alternative solutions. As our experiance grows, we become more aware of the possibilities. I don't think any of our class still uses TRIZ in their professional work, but learning the process of seeking inventive solutions from other applications is critical.

A parallel is perhaps biomimetics [1], an active current research area of 'design inspired by life/ biology'. Often, it is controversial if the invention was truely inspired by nature, or just made for good marketing. 'Shark skin' swimsuits were named by marketers after the actual invention [2]. Our experiance with heavier-than-air flight was terrible until we stopped trying to mimic how birds fly.

[1] https://en.wikipedia.org/wiki/Biomimetics [2] http://www.independent.co.uk/news/science/swimsuit-is-not-li...

Had bionics (the older name of biomemetics) on tip of the tongue for analogy sake too, but felt am not familiar enough with it. I however worked and shared hobbies with a few people who were into TRIZ: our alma mater was one of the hubs. Never seen it make a difference, but maybe it has its merits as pedagogic tool.

This somewhat reminds me of the "symbolic AI" software before the first AI winter in the early 90s... It looks like the methodolgy is making progress towards a solution, and you are able to find answers for heavily stylized sample problems...

...but then you realize a lot of the complexity is just hidden behind your definitions for the words in the TRIZ matrix (or in the case of "symbolic AI", in the precise semantic meaning you are assuming for the symbols)

You should check out the idea of function structures. Functions structures are a symbolic means of representing the functions a electromechanical product must perform in order to do its job.[0] The cool thing about function structures is that you can go from a function structure to a simplified conceptual model of a product. For example, one can replace the function convert electrical energy to mechanical energy with a motor. Unfortunately, the same function structure can translate to many different possible products, some of which don't make any real world sense.

For instance, for a carrot slicing machine you might end up with a solution that uses the forward motion of the whole device in conjunction with an airfoil to create the force to cut the carrot. While plausible, this is not the best solution.

If some of these problems could be fixed, we'd essentially have a hardware description language for mechanical engineering. And with this we could do the same sorts of design automation the semiconductor industry has been doing except for mechanical systems.

[0] http://enterprise.mie.uic.edu/me250/files/1113/4688/4799/woo...

ТРИЗ is awesome, as well as Altshuller's sci-fi writings.

nice

Reply on news.ycombinator.com