Thanks to Mad Men, the concept of brand equity has been widely popularized, the most prominent example being that of a dog food company’s brand becoming toxic after it was reported that it used horse meat in its product (a practice that was outlawed in the US in 1970).
D-Wave is still in the early stages of building its brand, so the fact that some casual observers have very negative views, as apparent from some comments on my last entry, is worrying. Unbeknownst to me, Scott Aaronson put up a blog entry about the same time and I didn’t immediately pick up on it as I was travelling. Turns out, the large comment string points to a D-Wave 2007 event as the cause for this brand damage. At that time the expectations of the theoretical computer scientists collided head-on with the reality of D‑Wave.
The comment thread at Scott’s blog is chock full of crazy, running the gamut from a poster accusing D-Wave of fraud to another one showering Scott with profanity. Always found it admirable that Scott doesn’t censor such comments but lets them speak for themselves (although Joy Christian proved that his saintly patience is not inexhaustible).
Scott initially challenged D-Wave strongly after the 2007 event (long before I started to pay closer attention – I heard about the 16 qbit claim at the time but didn’t consider such a chip yet commercially viable). Recently he buried the hatchet, but of course for a theoretical computer scientist the question of what kind of complexity class a certain hardware design can address will remain in the forefront.
Fortunately Daniel Lidar who is currently studying a D-Wave machine reported in the comment section that he plans to soon publish on exactly that. While other theorists such as Greg Kuperberg seem to be unwilling to move forward and past the 2007 “Sudoku crisis” (video of this “scandalous” demo can be found here), Scott and Daniel will clearly move this to a constructive debate based on published research.
The most prominent argument against this demo was that 16 qbits could not contain the entire parameter space of a Sudoku solution. Of course this ignores how the D-Wave system is implemented, and that it works in conjunction with a classic system as a kind of quantum coprocessor.
The Sudoku solver is not currently given as a coding example on the D-Wave developer portal. Since this old controversy can still cause damage to the company’s brand I think it may be a good idea to include it there.
It’s hard to gauge the extent of the brand damage that D-Wave suffered. Most IT folks will probably not have heard of the company yet, so this mitigates the problem, but on the other hand this left its traces in Wikipedia. The latter will almost certainly be one of the first stops for anybody new to the subject when trying to form an opinion. It is the first non-affiliated site that comes up in a Google search (when discounting recent news headlines) and it doesn’t get better from there. The next search result is this outright defamatory article from the highly respect IEEE, an organisation that is supposed to be objective.
Although my professional life is mostly IT consulting, I had my share of high tech marketing challenges to deal with (for a while as BI software product manager). In my experience a problem like this needs to be addressed heads on. My advice would be to very openly embrace this old controversy and to position how and why D-Wave’s design differs from universal gate-based quantum computing. The company does that in various places of their web presence, but it took me 1o minutes of googling to find this old physicsandcake posting that addresses this nicely in simple terms. A potential customer coming across old and new misinformation should be able to very quickly find these kinds of counter arguments.