Printing a bad story is a lot cheaper than launching a bad product.

To put it in a more explicit (but less succinct) way: The industry in which you are launching ideas will play a large role in how good your ideas need to be. And will therefore affect the process required & used to get to the final marketable ideas.

Recently, I've been mulling over what it takes to get more ideas out into the world: actual products & services in use. There are a variety of phases and it seems that different organizations get there in different ways. At this point, we don't have "best practices" for delivering innovative ideas/products/services.

I agree with the main point of Bob Sutton's blog post (where I snagged the above quote): you have to have "a lot of dumb, lousy, and crazy" ideas to get the good ideas - the ones that will be marketable and turn into end products.

But I hadn't yet considered the point of the quote: different markets have different costs for new ideas1. That concept is quite simple, but it is important to work out in practice.

I'm reminded of Joel Spolsky's article on the different approaches required for different kinds of software. He mentioned "Embedded Software" as one of 5 types of software. Software that is embedded on GPSes, cars, or fridges can only be written once. As opposed to software on your computer where you might get updates periodically. Or software on a website - that gets updated more regularly. Embedded software you have to get right the first time. Meaning the quality of ideas that goes into it has to be much higher. There is a higher cost to implement (more testing, better quality programmers needed, etc.).

The ultimate output of software affects the approach needed. The ultimate output market of ideas affects the approach as well.

Applying this view brings to mind a few specifics of differences between innovation high-cost marketspaces & innovation in low-cost marketspaces:

1. Leniency. Sutton's post references a 3% vs. 0.3% success rate of ideas for Onion articles vs. Mattel (& others) toys. The higher the implementation cost (& time), the higher the bar for ideas. You can't play around as much.
2. Prototypes: Number & Quality. At the onion, the prototype is very lo-fi: 2-3 sentences on a whiteboard. For ideo, they start with ideas on a board, then move to drawings, then possibly move to high-fidelity working prototypes.
   The higher the implementation cost, the more you spend time & money on trying progressively larger prototypes.
   For a website, they'll just drop a new feature immediately: it may have been verbally discussed with the team a few hours before2.
3. Testing. This is inversely related to prototypes. If cost of implementation is cheap then you can change course quickly. A "throw it out the door and iterate" approach is do-able. This is usually done in conjunction with testing. For example: you push a new feature to a portion of users of a website and test the results against a control group to compare against.

I'm sure that there are more differences. But taking a look at the market in which you are trying to innovate will tell you something about the approaches that are appropriate.

Notes

1. As usual with many of my favorite thoughts: this one was a sideline or throwaway comment that was symbiotically clinging to the main thesis.
2. There are many examples of websites pushing features out the door quickly. A recent video interview with Paul Buchheit among others (he's the founder of FriendFeed and previous created GMail) has a great nugget where the Scoble (the interviewer) asks how they test new products and Buchheit's response is that they basically just push things to the website instantly.
   (Trying & failing to find this quote in the video reminds me of the VCR & information overload thoughts.)
   http://www.fastcompany.com/scalability Note: email address required to watch the video.