Wikinomics by Don Tapscott and Anthony Williams
Despite the name, which suggests both a me-too jumping on the Levitt/Dubner Freakonomics bandwagon and a possible reductive identification of all evolving Web technologies with wikis, this is a surprisingly good book, written at a calibrated level of abstraction, with a tasteful blend of concepts, anecdotes and statistics. It has none of the anecdotes-of-a-gunslinger-economist machismo of Freakonomics, and the wiki in the title is synecdoche, not reductive imagining.
I read the thing in one sitting at Barnes and Noble a few months ago, taking notes and annoying the staff (I conspicuously displayed my coffee cup to show I was legitimately avoiding buying the book — I now have a preprint copy via a colleague who knows Don Tapscott personally). That I could do so tells you that the content was pretty light reading. It was concise and valuable though, and about as deep as you can reasonably expect a first-to-market book on the topic to be. It’ll be a few years before we see the dense, canonical and empirically-validated book we all really want, so until then, this book is a pretty good operating manual. You cannot divorce the themes of the technology of virtualization and the culture of open innovation, and this book proves the point. It is as much about open innovation as it is about things like wikis and blogs.
Key Points
Few authors summarize their key points comprehensively in the introduction, so I always appreciate it when a writer does that. You can actually get away with reading just the introduction if you want 80% of the value and are willing to accept the assertions as best-faith truths. For good measure, the conclusion reiterates the same points with some added depth. The summary is presented in the form of a few “principles of wikinomics” that almost constitute a doctrine:
- Being Open: maintaining porous organizational boundaries, working with industry standards, moving to a culture of non-secrecy, above and beyond legal needs, and being open to globalization and global trade. Well argued.
- Peering: This uncomfortable neologism is meant to capture the themes of encouraging self-organization and working with non-hierarchical organizational structures where possible. Moderately well argued.
- Sharing: To move from a win-lose business-is-war mindset to a mindset where there are win-win/coop-etition options available. The key idea here is to think about business IP as a managed portfolio that includes things you give away (including to competitors) alongside things you protect with patents or hide with secrecy. This is a slightly more elaborate conceptualization of the IP market than Henry Chesbrough’s in Open Innovation
which I’ll get to.
- Being Global: This point is made a little obscurely, and I didn’t quite see where it fit in with the rest of the picture, but never mind. One must counter Tom Friedman’s annoying Flat World metaphor where possible, so this might just be called community service. There’s some resonance, so let’s leave it there.
- Generation-N Norms: The idea being that this whole culture shift is being driven by Generation N (his word for Baby Boomleters) norms, that include speed, freedom, openness, innovation, mobility, authenticity and playfulness. Yes, I too gagged a little at the primary-colored bubblegum utopia-ness of it all, but to be fair, the authors do justify the list throughout the book with substantive examples. I can just see skeptics translating that to sloppiness, lack of accountability, competitive carelessness, lack of execution discipline, ADD, disloyalty, political naivete and silliness. That set of dichotomies is a whole other post.
Highlight-Worthy Themes
Some highlights cherry-picked for your edification. There’s a chapter-wise summary after this section, for those of you who must have completeness.
Goldcorp
Rather smartly, the authors chose to start the book off with a success-story anecdote from a very traditional industry. This is the story of Goldcorp, a gold-mining concern in Canada. Goldcorp was faced with diminishing yields from its mines and an unclear idea about where to dig for more gold. The company’s traditional staff geologists failed to make much progress with their safe, ponderous way of exploring as they’d always explored before. The CEO made an incredibly bold decision to throw open the exploration problem to everyone, making all the geological data available to all (apparently unheard of in the mining industry, where this data is held very closely). An enthusiastic army of outsiders, ranging from computer scientists to mathematicians went to work (along with lots of outside geologists too, of course) and came up with a huge number of suggestions for places to dig, half of which the in-house staff hadn’t thought of. The company went from $100 million to $9 billion in the space of a few years (this was 2000-2004 I believe). Remember the famous story in The Wisdom of Crowds
Stepping back, the whole darn book has this flavor. Yes, there are chapters covering Linux and Apache and Wikipedia, but repeatedly the authors emphasize the principles and non-traditional examples. They clearly know who they are trying to pitch the ideas to, and it is NOT the Linux choir.
Coase’s Law
One of the new things I learned was Coase’s law, which can be informally stated for our purposes as “Organizations exist in the form they do because of the relative costs of information per transaction” (my wording). Specifically, every action that adds value on a product/service’s path to market involves a search cost (finding someone to do/make it), a transaction cost (the cost of making a deal with the producer via negotiations) and the actual cost of doing the work.
It is a neat phenomenological law to help think about the dynamics of creative destruction in the sense of Schumpeter. Traditional organizations are large not just to achieve economies of scale, but also to ammortize the first two kinds of cost over many instances of the third kind of cost. The reason this is important for Wikinomics? When the search and transaction costs plummet (think Google+eBay or eLance), tasks can drift around in the economy, in and out of porous boundaries, to find their least-execution cost homes. The implication is that traditional organizations are structurally irrational in the present day, and mostly need to shrink to achieve their right size to reflect the new cost-of-information dynamics. The leaking function gets out there into a brave new frontier of intermediate markets.
Global Plant Floor
Okay, so we’ve all heard the classic open-source success stories, ranging from Linux to the Wikipedia. Can a similar philosophy build physical products? Can you create an open-source Boeing 747 purely through a mob refining a design on the public Internet and using commoditized manufacturing shops for fulfillment?
Second question. Can the process truly invent new ideas? After all, the biggest open source successes are clones or minor variants of previous commercial products, not totally new ideas in software — yes wikis and blogs are cool ideas that could be said to have open origins, and yes, Richard Stallman did invent emacs, but still, there isn’t a compelling radical innovation story in the open source world.
The book offers little insight into the second question, but for the first question, the book provides a tentative answer: Yes. China’s motorcycle industry (Lifan) is offered up as one example (though it is a swarm economy that mainly rips off Japanese original designs — innovation without invention). The more impressive example is the Boeing 787, contrasted with the Airbus A380, which is a logistics and manufacturing planning disaster (and possibly a design integration and multicultural disaster as well). One way to assess the A380 is to call it a disaster of waterfall planning of the Windows Vista variety. Contrasted with this, the Boeing 787 dreamliner has been developed with a very carefully calibrated open-market approach where the entire supply chain of subcontractors has been brought into the design phase with a “the guys who will manufacture it should design it” philosophy. Boeing has only retained the critical-to-design features like the tail fin.
This would be an impressive example if the 787 were already in the market, beating the A380 hollow. We’ll have to actually see how that battle plays out.
The critical value drivers of the ‘global plant floor’, per the authors are:
- Orchestration (what Boeing is doing)
- Rapid iterative design processes (obvious for software, but hmm… maybe we don’t need a decade long gestation for complex hardware systems either).
- Modular architectures (the The Age of Modularity
is now on my reading pile)
- Build an innovation ecosystem (let’s all develop Amazon S3 clones)
- Share costs and risk with partners
Wikinomics Design Principles
Now this is another interesting list from somewhere in the book. I’ll defer a discussion to after I post about disruptive innovations and open innovation, since deconstructing this list requires analysis beyond the scope of a book review.
- Take cues from lead users
- Build critical mass
- Produce an infrastructure of collaboration (open standards, legal foundations)
- Make sure value is shared by all prosumers
- Abide by community norms (the N-gen ones?), including IP sharing.
Chapter Summary
After the introduction chapters, there are 8 chapters, each with a cutesy name. The overall structure is one of ‘Let’s collect the high concepts’ — don’t expect a business model recipe. In brief:
- Peer Pioneers: Covers Wikipedia, Apache and Linux. Some new points here, even if you’ve already heard these stories
- Ideagoras: Play on Pythagoras, I presume, is about open economy broker outfits like Innocentive (biochemical industry), Nine Sigma and a bunch of industry leaders. Some interesting statistics here: Proctor and Gamble discovered it used less than 10% of the patent IP it generated. Now it gets 50% of its IP needs from outside and cheerfully trades stuff it doesn’t use. IBM gets $1 billion from IP sales. More when I talk about Open Innovation.
- Prosumers: Includes coverage of Second Life, the “hack this product” story from Lego Mindstorms (they tried to close the API and sue hobbyists who tried hacking it, and then realized that was actually a good thing and opened the product up). Steve Jobs’ iPod is in there too, as a story with mixed lessons.
- New Alexandrians: Is an attempt to get at the research culture of Wikinomics, and there is a discussion of Intel’s university-lab strategy, as well as arXiv (the physics preprint clearing house), MIT’s open courseware, Merck releasing 15,000 formulas and the human genome project. This is a weak chapter — there is a rather forced air of trying to fit too many not-quite-appropriate examples into the conceptual framework.
- Platforms for Participation: This covers the open-API+innovation-ecosystem-around-us represented by Amazon web services and the Google API. Also includes a discussion of the crowd-coordinated work in the aftermath of Katrina and the opening up of that famous piece of closed software, SAP.
- Global Plant Floor: I covered this as a highlight, so let’s ignore this one.
- Wiki Workplace: The money chapter for most of cubicle dwellers, but with surprisingly little value for understanding how the workplace is changing (either for workers or management). You want to understand wikis and blogs in your daily world, use them there and find out for yourself. Books full of ambitious abstractions is the wrong way to learn that game.
- Perfect Storm: Ties the whole book together and weaves all the themes into a narrative of looming seismic shifts in culture. Hard to summarize this chapter. Suffice it to say, this will be the watershed chapter. You’ll either walk away saying the chapter is too timid in its predictions, or believe it is all hype.
The book has its own website, with some sample chapters, in case you want to get the story straight from the source.
About Venkatesh Rao
I read about the 787 in Friedman and cringed. While Airbus has made some spectacular mistakes, I can’t help feeling that Boeing has taken a much harder route. Imagine the interfacing and integration issues when you outsource the design and implementation of different parts to Japanese who re-outsource them to Russians. I mean, the language, culture, units of measurement… ripe scenarios for “Galti se mistake ho gaya” ([tr: “By mistake we have made a galti”).
And the reverse auction, where Boeing sets a fair price for, say, a bolt, and suppliers bid against each other to drive the price down to presumably cut-me-own-throat levels. I’m sure they’d have to meet quality criteria and all that, but without continuous and effective policing of quality by Boeing, the lowest bidder would have a lot of incentive to cut corners.
I don’t know about you, but when all that’s keeping my body and hypothetical soul from becoming one with the Wild Blue Yonder is a thin shell of metal and plastic, I *want* all the corners as uncut as possible.
I guess I’m so paranoid because a large percentage of such divide and conquer efforts in software are disasters. Who knows, maybe Boeing can do it. Then they can give it a nice sporty name, call it a development methodology and start selling books and seminars to gullible managers in software companies who are always on the lookout for the next fad diet which will make them lose 20 kg *and* become more agile.
I was intrigued by your re-statement of Coase’s law. The Wiki article on Ronald Coase explains it well – the section about “The Nature of the Firm” is the most relevant. http://en.wikipedia.org/wiki/Coase
tubelite:
Good points about the 787. the thing is a sales miracle and a case study in open innovation and ‘open source’ style manufacturing, but unless we go a decade without the thing falling apart in the sky, we won’t know for sure. Airbus’ competing product (which isn’t the A380, but the A340WB) is also similar engineering wise (composites based), but the manufacturing model isn’t so radical. But in a way, you might say Boeing has been forced into this model partly by the demonstrable lack of success of Airbus’ model which has produced a revolving door parade of CEOs in recent years. The latest Economist has a piece on Airbus finally turning the corner.
Incidentally, it is interesting that you should pick translation errors as a potential flaw of Boeing’s open model. French/German errors were one big part of the A380 mess, so clearly that particular failure mode applies to both open and closed development models.
Yengul: my restatement is really a paraphrase of Tapscott/Williams’ use of Coase’s law. I did look at the Wikipedia articles, but then decided to do a deep dive into Coase for a later post — am currently reading the original ‘The Firm, The Market and the Law.’ If I can get up to speed on the concepts, I might do a post on Coase’s law and its relation to information theory later. Complex idea.
This notion of “open source” design of mechanical products is not really all that radical. The adoption of “design+manufacture” outsourcing by the aerospace industry is just a natural progression from what is already happening in its sister space – auto. Toyota has, for some 25+ years now, used a system where they will first spec the “hard points” of a platform (e.g., the tentative locations of the wishbone strut pins, or the expected top coordinates of the front suspension damper, or the rear axle width etc., as well as expected gross weight, max speed, horsepower etc.) with some tolerance ranges, and then invite a select group of suppliers to commence parallel tiger teams on designing the subsystems that will “anchor” onto these hard points (say for subsystems like a front suspension, for an HVAC system etc.). So when the blueprints arrive from the suppliers, they come with tentative BOMs as well (quotes, based on some projected volumes). Toyota engineers now can not only trade off on this set of options by price, they can also trade off in terms of dynamics, weight, safety etc. The winning design is then sourced primarily from the winning vendor, and maybe farmed out to competing contractors as well (for volume, like the Pentagon does with rival warplane platforms after a bake-off) – but there is no “Reverse auction to get the vendors to cut their own throats”. There is actually a significant amount of trust and caring between toyota and the qualified vendors who are within its system. Toyota can thus free up its own bandwidth to focus only on the “crown jewel” elements that take the best of its heavy engineering – powertrain stuff mostly. The design+sourcing system is better because it forces the responsibility of design away from the OEM, and into the hands of the supplier, pushes out creativity and accountability deeper into the value chain, and enables a greater set of tradeoffs to be examined by an effectively larger pool of talent (both toyota engineers, as well as design teams at 3-4 subcontractors will evaluate design options for a subassembly). The OEM can then spend more of its time on core and forward-looking technologies, and on its real strength – sales & marketing. This is EXACTLY what Boeing is doing with the 787 in many ways. This approach is not equally prevalent everywhere – some US OEMs are further along than others – I believe Chrysler was (though its sourcing engine could not really get enough scale over its smaller volumes, so the company ended up having the theoretically “best” designs, but practically worst execution), and in Europe, the smaller ones like Peugot/Citroen have been furthest along in pushing out the burden of design onto their suppliers, while they retrench into marketing and core-future-platform dev work only. This is finally starting to hit the truck business globally – since this used to be a regionally more siloed market to date, with a heavier focus on core powertrain R&D than on the other parts of the vehicle – but with globally converging emmissions standards over the last few years, the pressure to outsource more and more of each platform to a set of globally qualified suppliers is hitting most OEMs. There are some other open-source hardware design projects – notably the Nicholas Negroponte OLPC program (that just partnered with Intel yesterday).
In general, the one factor that HOLDS BACK open sourcing in hardware is the non-zero marginal cost of a hardware part. In software, you write once, and then repeatedly reuse components in both future designs, and in actual future implementations. While opensourcing may indeed apply more broadly to the “design/architecture” part of hardware creation, it is much more difficult to align economic interests of a supply chain around it for the ongoing manufacture. Some of this happens naturally at a “dumb component” level – e.g. with screws, fasteners and other standards – but it is externally imposed (to reduce redesign and rework) and not really developed grass-roots as an alternative to ground-up OEM designs that embed fat margins. That is one of the main reasons an OEM cannot simply have its suppliers take on the burden of detailed design, and then turn around and screw them on the marginal cost as well. I think European vendors who have started outsourcing parts of their systems have applied this kind of brute force squeezing (because the OEMs themselves are often in small, fragmented, sluggishly growing markets with intensely rapid fasion trends). Toyota has built enough trust and care into its supply chain globally to make this work though – and manages to hold a cost position more robust than any peers’!
I wonder if there will be more open source development as mechanical complexity gets over-ridden by the electrochemical complexity of vehicles in the future. Many more components in the car of the future (batteries, buses, motors) will be shareable across OEMs under standard descriptions, than the mechanical systems we see today.
Torp:
Good points all, I particularly like the point about building trust in a network of suppliers and not creating an anarchic bloodbath of cost-cutting. But remember, the open source doctrine is about more than cost. It’s key tenet is “with enough eyeballs, all bugs are shallow.” Would be nice to see that applied to a hardware situation. Hard, since hardware testing is massively expensive.
I am not sure I buy your major conclusion about the incremental cost of h/w being the main bottleneck. When manufacturing in a sector can be truly commoditized, then it can become open source. The author of “The 4 hour work week” tells of how he managed to get a seamless value chain down to China.
The semiconductor industry is an interesting example. At the high end, Intel needs to clone entire factories, down to the janitor’s closet, to capture a successful model of processor fabrication. But lower in the food chain, most ASICs, memory chips etc., are designed by design houses and manufactured elsewhere by commodity fabs. The fabs themselves seem to go through a cycle of shake-ups in sync with the business cycle. This is ripe for open sourcing. I am just waiting for a smart fab in Hong Kong to accept open source design fulfilment. Margins are razor-thin and fab utilization within its lifetime (driven by vibration constraints from structural aging) has to be near-perfectly calculated for the fab to survive, otherwise it dies in the next business cycle downturn.
If fabs figure out a way to do flexible manufacturing of very small lots of chips (I think they work off discs of silicon about 12″ diameter) and can accept chip designs submitted over the Internet (maybe via a sourcefourge like open source repository for chip designs), then there might be some serious money there. There is a huge move away from ASICs to FPGAs, my hardware friends tell me, and maybe this can be reversed with the right open source model.
Amazing what a difference a year makes… to the 787… see http://snipurl.com/4e60t
The isue seems to be that when an industry takes a step onto a different ‘s’ curve for learning about new technologies (tools, techniques, materials and methods) it stumbles.. as a design engineer I went through the same when the big-fan engines and Tristar/DC10, B747/Airbus were being introduced… how traumatic that was for all involved… it is little different this time.. just faster feedback!
That “ideagoras” word is not a Pythoras reference, but “agora” reference.
http://en.wikipedia.org/wiki/Agora