Mindstorms NXT: Mindstorms Resurrected?

Lego's announcement of Mindstorms NXT an upgrade to its Mindstorms line of computer-controlled/programmable construction toys has generated a tremendous amount of excitement amongst Lego fans. The announcement has been especially welcome as Lego had forgotten about the Mindstorms line since the introduction of the Robotics Invention System 2.0 (RIS2) construction set back in September 2001.

Computer-control as introduced by the Lego Mindstorms product line is arguably the biggest innovation in construction toys in decades. Computer-control turns mundane construction toys into "intelligent" autonomous robots that can interact with the environment in interesting ways. It allows construction toys to recognize and respond to stimuli and enables entirely new possibilities for learning and play. Most importantly, by lending sophistication to construction toys, computer-control extends the appeal of construction toys to a broader audience. Disappointingly though, Lego has shown little interest in selling computer-controlled construction toys.

The computer-controlled construction toy evolved out of Lego funded research at MIT Media Lab. Lego commercialized the concept in September 1998 with the $200 Robotics Invention System 1.0 (RIS1) construction set, which became an instant hit with techies and geeks. Subsequently however, Lego never issued a comprehensive update to RIS1; RIS2 was only a minor update to Robotics Invention System 1.5 which itself was a minor update to RIS1. Moreover, Lego released the RIS2 without fixing compatibility problems with Windows XP, the mainstream operating system of the time.

With Mindstorms NXT, Lego has finally decided to address the compatibility problems of the RIS2, but amazingly, Lego is still uninterested in selling computer-controlled construction toys in volume.

Construction toy shipment volumes are highly dependent on price. Lego's chief competitor in the construction toy business Mega Bloks is succeeding against Lego primarily on the basis of price. Mega Bloks toys are regarded as low quality in comparison with Lego toys, but Mega Bloks' revenue grew from $114.0 million in 2000 to $234.6 million in 2004. Mega Bloks' 2003 annual statement says that over 90 % of Mega Bloks' products retail for $29.99 or less, and its 2004 annual statement says the company sold over 30 million toys in 2004. Clearly, Mega Bloks is posting impressive revenue growth numbers by selling inexpensive construction toys in very large volumes.

Lego practically owns the market for construction toys priced $30 or more, but Lego prices very few of its toys beyond $90. The Lego Users Group Network (LUGNET) maintains a fairly comprehensive record of Lego construction sets/toys that have been released to date. The LUGNET database lists 303 new Lego sets as being released in 2004 and has pricing information on 224 of them. Of the sets with pricing information 62 were priced $30 or more, 21 were priced $60 or more, and only 8 were priced $90 or more with four sets priced at $90 and the most expensive set priced at $140. Evidently, beyond $90 there is not much of a market for construction toys, and Lego is earning most of its revenues by selling relatively inexpensive toys in large volumes.

Lego priced the Robotics Invention System (RIS) at $200, and at that price it could not have sold well. Not unexpectedly, the RIS did not sell too well. According to numbers disclosed in a recent Wired article, Lego shipped a total of 1 million RIS units in all. This sales volume is not at all impressive as Lego shipped the 1 million units over a period of seven years, and just Mega Bloks alone shipped over 30 million construction toys in 2004. Moreover, Lego not only mispriced the RIS but it also mispriced other standalone Mindstorms sets.

In September 1999, one year after the launch of the $200 RIS 1.0, Lego added two new sets to the Mindstorms line: the $150 Robotics Discovery Set and the $100 Lego StarWars Droid Developer Set. The Robotics Discovery Set and the Droid Developer Set differ from the Robotics Invention System in the amount of programming control they offer to the user. While the RIS is fully programmable by the user, the Robotics Discovery Set only offers the user limited programmability, and the Droid Developer Set is not intended to be programmable by the user at all. Like the RIS, the Robotics Discovery Set and the Droid Developer Set could not have sold well to regular Lego customers because of the premium pricing. In fact, Lego subsequently discounted these sets heavily in order to dispose off inventories.

Interestingly, with Mindstorms NXT, Lego is persisting with its strategy of premiumly pricing computer-controlled construction toys. Instead of pricing Mindstorms NXT low in order to enhance sales, Lego is doing exactly the opposite. Relative to the RIS, Lego has raised the price of Mindstorms NXT by $50 to $250. Consequently, Lego can not reasonably expect to ship very many Mindstorms NXT sets.

An earlier Techuser article, Lego Mindstorms: What Went Wrong, examined the reasons behind Lego's reluctance to introduce computer-control at an affordable price point. That article claims that consumer interest in the Mindstorms line is high and Lego's profit margins on Mindstorms sales are good, but Mindstorms sales cannibalize Lego's traditional construction toy sales, and this issue is "forcing" Lego to neglect computer-controlled construction toys.

Unfortunately, the analysis in the above-mentioned article is incomplete. Lego in fact made an attempt to sell large volumes of computer-controlled construction toys. The attempt, however, did not work out well for the company, and since then Lego has practically given up on computer-controlled construction toys.

In September 2002, Lego introduced the Spybotics line of four computer-controlled toys/sets priced at $60 each. The Spybotics line was based around the Spybotics electronic brick. The Spybotics brick contained a microcontroller, two motors, two sensors, and IR communication circuitary. The brick was completely end-user programmable; although, the Lego provided PC software only allowed for downloading of a limited number of predefined programs to it. For the $60 price, Lego even threw-in an IR remote control, an indispensable accessory for controlling computer-controlled toys that Lego never bundled with the RIS.

The Spybotics sets were designed to assemble into vehicles called Spybots. The Spybots could be controlled via the IR remote control but could also act autonomously under computer-control. The Spybots were able to sense and react to light sources, physical obstacles, and other Spybots.

The Spybotics line was squarely aimed at taking back the ground construction toys have lost to video games. Play with the Spybots involved completing ten Lego provided real-time strategy game style missions with the Spybots acting as real world equivalents of strategy game combat units. The missions came on a PC CDROM and were downloaded to the Spybots via a serial port connection and a CG heavy video game style interface.

Lego took considerable pains to price the Spybotics sets at $60. To hit the $60 price point, Lego dispensed with all printed documentation in the Spybotics sets, including Lego's trademark printed assembly instructions. Instead, Lego bundled all documentation in electronic form on a CDROM. Additionally, Lego employed the obsolete RS232 serial port in Spybotics sets for PC connectivity. This move could only have been aimed at saving pennies on the cost of the microcontroller in the Spybotics brick as the serial port was obsolete even back in September 2002; by that time, even Lego had moved on to USB in the RIS2. All of this suggests that $60 is the price point beyond which construction toys do not sell well, and Lego chose the $60 price point as it wanted to ship large volumes of Spybotics sets.

The Spybotics sets had a lot in common with the Droid Developer Set. Like the Droid Developer Set, the Spybotics sets were crippled versions of the RIS sets with actuators/motors and sensors integrated in the basic electronic brick. Additionally, the Spybotics sets too only allowed for the execution of a number of predefined programs.

This connection between the Droid Developer Set and the Spybotics sets is not coincidental. MIT Media Lab developed the technology underlying the Mindstorms/Spybotics product line, and a Media Lab paper, Mindstorms: The Structure of an Engineering (R)evolution, provides insight as to Lego's overall Mindstorms strategy. The paper states:

While the ideal scenario would have LEGO market its [Mindstorms] product (and its philosophy) to all children regardless of age or gender, the necessity to realize a return on investment drove a decision to market to a more specific segment. At the same time, executives at LEGO felt that the toy company could hit a public relations "home run" if it identified a target market and outsold expectations.

Once a decision was made to limit the target audience, the choice of 10-14 year old boys was an obvious one. This market segment has long been LEGO’s "sweet spot." Additionally, market research done by the toy company [Lego] indicated that boys would be more attracted to "computerized" toys than girls would be . . . .

The decision to market to 10-14 year old boys brought with it several influences over brick design. Namely, the choice of colors (primarily yellow and black, which mimicked a construction zone) [26] and the type of sample applications (e.g. robots, etc.) [21] were directly related to the target market.

Basically, the goal of the Mindstorms line was market research: Lego tried to identify a target market for computer-controlled construction toys. Lego wanted to use "computerized" toys to grow its revenues by selling to individuals who were not buying Lego toys. Lego used the Mindstorms line to see if "computerized" toys will sell to 10-14 year old boys who had moved on to video games, and additionally, it used the Mindstorms sets to evaluate the appropriateness of different formulations of computer-control for the target market.

Lego eventually settled on the formulation of computer-control in the Droid Developer Set, and created the Spybotics line to introduce computer-control to the target market. Lego chose a "programming free" formulation of computer-control as programming demands a fair bit of instruction, and Lego did not want programming to become a hurdle in the way of mainstream adoption of computer-control.

In keeping with its goal of shipping large volumes of Spybotics sets, Lego aggressively marketed the Spybotics line. An article by the marketing portal ClickZ suggests that Lego had a six-figure budget for Spybotics' UK launch alone. Obviously, Lego had big expectations from the Spybotics sets and was hoping to sell Spybotics sets in the millions.

Unfortunately, contrary to Lego's efforts and expectations, the Spybotics line did not fare too well. By early 2004, Spybotics sets were being dumped at $29 a pop at Fry's electronics. In fact, Lego had given up on the Spybotics line by early 2003, merely six months after the launch of the Spybotics line.

The Spybotics website was designed as a repository for new Spybotics missions and was central to the Spybotics experience. The Spybotics website, however, only lists two Spybotics missions. The missions are in the form of zip archives with modification dates of 12 December 2002, and 7 February 2003. These dates suggest that Lego gave up on the Spybotics line right after sales figures from the 2002 holiday shopping season came in.

Lego not only discontinued the Spybotics line, but subsequently refrained from attempting to sell computer-controlled construction toys in volume. A Kidscreen Magazine article, Tech toys are fighting an uphill battle for shelf space, sheds some light on what went on. The article states:

In 2002, Lego introduced its Spybotics line of interactive robots (US$60) that can play simple games like Capture the Flag and Laser Tag. "When we had the opportunity to show the robots in action, people were very impressed," says Jeff Ardis, Lego's brand manager for tech toys. "But it's difficult to communicate on the shelf exactly what the product does. In some cases, we couldn't get around that."

When mass retailers like Wal-Mart placed the Spybotics products with other Lego SKUs, the sell-through was much lower than anticipated. Toys 'R' Us was able to find a place for Spybotics in its 'R' Zone interactive gaming section because shoppers there were already in the mood for products with a tech bent. Ardis says sell-through jumped up about 35% in Toys 'R' Us compared to stores where the Spybotics were merchandised by brand.

However, TRU recently decided to move the line back into the toy aisle, says VP of merchandising Wayne Yodzio. Because the 'R' Zone is really about video games and the margin is so much lower on tech toys, he says the space required for the Spybotics didn't justify their existence in the section.

The Kidscreen Magazine article seems to suggest that Lego abandoned the Spybotics line because it never figured out how to market it. However, by the time Lego abandoned the Spybotics line, Lego had figured out how to sell the Spybotics sets. Lego's Toys 'R' Us experiment of putting Spybotics sets in the 'R' Zone had revealed that the company needed to repackage the Spybotics sets and sell them alongside video games.

Lego needed to shrink the dimensions of the Spybotics boxes in order to get retailers to place the Spybotics sets on video game shelves. A typical Spybotics box measured 15 x 11 x 3 inches and occupied 495 cubic inches, while a computer game box typically measures only 10 x 9 x 1.5 inches and occupies 135 cubic inches. Not only were the dimensions of Spybotics boxes completely inappropriate for shelves intended for computer games but they were taking 3.6 times the shelf-space of the average computer game package. Console games come in even more compact packages, so the Spybotics boxes could not be placed near them either.

The Spybotics boxes were mostly empty space and Lego could have easily shrunk their dimensions. Lego was using oversized boxes not because the Spybotics components were oversized and could not fit inside compact boxes, but because Lego originally intended to sell the Spybotics sets in the construction toys section. In the construction toys section people associate oversized boxes with value and shelf space is not that big of an issue.

Lego never changed the packaging of the Spybotics sets to incorporate what it learned from its Toys 'R' Us experiment. Evidently, low sales volume was never the real problem. Low sales volume could not have been an issue for another reason: many Lego product lines failed to sell well in the period after the launch of the Spybotics line; however, Lego did not discontinue those product lines. Lego's 2003 annual statement states:

At the beginning of the year, substantial inventories of LEGO products, particularly in the USA, had built up within the retail channels as the 2002 Christmas sales had failed to meet expectations. As a result, sales were extremely slow at the beginning of 2003. . . .

The decline was particularly marked in the US where sales dropped by approx. 35 percent compared to 2002 while Asian markets experienced a fall of 28 percent. A significant part of the decline in these markets can be attributed to shortfalls in the sale of movie tie-in products.

Lego had invested more than four years actively investigating the potential of programmable construction toys and a minor marketing setback could not have dissuaded Lego from trying again.

The real reason for Lego's retreat had to be something different and it was cannibalization. In its 2003 annual statement, Lego admitted to cannibalization being a problem for the company. The annual statement says:

For several years, LEGO Company has invested substantial funds in expanding its product portfolio. This commitment and the consequent cost increases have not produced the desired results. In some cases, new products have even cannibalised on the sales of LEGO Company’s core products and thus eroded earnings.

The earlier Techuser analysis had suggested that the cannibalization was due to the Mindstorms line, but it was the Spybotics line that caused Lego problems.

Lego had expected the Spybotics sets to sell to people who typically do not buy Lego sets, but instead, the Spybotics sets ended up selling primarily to regular Lego customers. Lego ended-up selling low-margin Spybotics sets to people who would ordinarily have bought high-margin Lego sets, and in the process also ended-up with unsold inventories of the high-margin sets. Worse, the Spybotics sets with their fancy play redefined the Lego experience for people who purchased the sets, and such individuals became reluctant to purchase plain vanilla Lego sets.

Lego realized that to address the cannibalization it will have to pervasively integrate computer-control in many of its product lines in order to provide sufficient choice to individuals who loved the Spybotics experience. Such a move would have cost Lego a lot of development effort as well as lower margins on Lego sets, and Lego took the easy way out and chose to bail out of computer-controlled construction toys.

The attempt to sell Spybotics sets in video game aisles was only a response to the cannibalization. Lego was hoping that in the video game aisles regular Lego customers will not buy the Spybotics sets, but apparently, the move did not help.

Lego's basic miscalculation was that it extrapolated the sales patterns it identified in Mindstorms sales in order to put together its Spybotics sales strategy. Lego sold unexpectedly large volumes of Mindstorms sets; it sold Mindstorms sets as construction toys; and it sold Mindstorms sets primarily to people who were not regular Lego customers. From these patterns Lego inferred that Spybotics sets will sell well as construction toys and they will sell primarily to people who are not Lego customers. Of course, Lego got everything completely wrong as the Mindstorms' sales patterns were a consequence of their overpricing and the media's fascination with everything robotics.

The overpricing of the Mindstorms line had the effect of turning away regular Lego customers. Ordinarily, the overpricing should have depressed sales to miniscule levels. However, the media overhyped the Mindstorms sets because of their robotics orientation, and this led to a flurry of purchases from people who would not have otherwise purchased Lego sets. The net result of all of this was promising Mindstorms sales with no significant cannibalization.

Unfortunately for Lego, the Mindstorms sales patterns did not materialize for Spybotics sales. As soon as Lego priced computer-controlled construction toys more reasonably, traditional Lego customers started buying such sets. Moreover, the media did not cover the Spybotics line as extensively as it had covered the Mindstorms line and mostly regular Lego customers bought the Spybotics sets. The net result in the case of Spybotics was cannibalization of traditional Lego sales.

Lego's Mindstorms NXT announcement suggests that Lego has decided to refrain from introducing computer-controlled construction toys at a mainstream price point for the foreseeable future. The Mindstorms NXT infrastructure does not lend itself for commercialization at a mainstream price point. Lego is using a 32 bit ARM microcontroller with 64 KB of RAM and 256 KB of Flash memory in the Mindstorms NXT programmable-brick. Such a microcontroller costs somewhere around $8, and is clearly not appropriate for use in products intended to retail for $50-$60. Lego might be able to get away with using a lower-end 32-bit microcontroller that costs around $3, but even $3 is a substantial cost for a component that is going to go in a cheap toy. On the other hand, 8-bit microcontrollers, like the ones used in the older Mindstorms line, cost significantly less than 32-bit microcontrollers, and if Lego intended to introduce programmable construction toys at a $50-$60 price point, it would have stuck with a design based around an 8-bit microcontroller.

The other interesting thing about Mindstorms NXT is that it is not much of an upgrade. Experienced model builders do not demand fancy sensors and actuators; they mostly want to be able to utilize lots of sensors and actuators to give their models sophisticated behaviors. The RIS allowed for interfacing with three actuators and Mindstorms NXT has not raised this limit. This limit reflects Lego's insistence on powering the Mindstorms NXT by six AA batteries. Six AA batteries do not have enough juice to adequately power four motors of the kind that come with Mindstorms NXT, and this is why Lego is sticking with only three actuator ports.

The Mindstorms NXT update makes sense only in the context of the educational market. In the educational market Lego can not bail out of supplying computer-controlled construction toys. Companies like FisherTechnik are already supplying programmable construction toys to that market, and by bailing out Lego will only cede the market to its competitors. With Mindstorms NXT Lego is attempting to enhance the competitiveness of its offerings by issuing a major update to the Mindstorms software and is also simultaneously addressing the cannibalization problem.

Cannibalization has been a problem for Lego in the educational market as Mindstorms sets offer a much richer experience than regular Lego. This makes school kids perceive regular Lego sets as inferior and makes them reluctant to buy regular Lego sets. With Mindstorms NXT, Lego has come up with an ingenious solution for this problem. Lego has gone totally studless with Mindstorms NXT. By getting rid of studs, the most distinctive feature of Lego blocks, Lego is hoping that school kids will not be able to identify Mindstorms NXT pieces as Lego. The expectation is that kids will see regular Lego and Mindstorms NXT as two different types of toys and not as inferior and superior versions of the same type of toy.

So far Lego has done a decent job at damage control, but long term such a policy will be suicidal. The cannibalization of regular construction toy sales by computer-controlled construction toys clearly suggests that computer-control is a disruptive technology in the construction toys market. Lego has managed to keep computer-controlled toys under wraps only because it is practically a monopoly in the construction toys business. To be able to sell computer-controlled construction toys to the mass market, Lego's competitors need to have a decent assortment of inexpensive mechanical construction parts for building models, substantial expertise in manufacturing computer-controlled construction toys, and good distribution and brand awareness. None of Lego's competitors have all the prerequisites. For instance, FischerTechnik has the first two prerequisites but lacks the distribution and brand awareness to sell to the mass market.

Interestingly, the biggest threat to Lego comes from companies who are not in the construction toy business at all. If programmable construction toys takeoff, the big winners will be Nintendo, Sony, and Microsoft. These companies will win because they make (or want to be in the case of Microsoft) handheld gaming devices and such devices are the perfect controllers for programmable construction toys.

Current generation gaming handhelds come with wireless connectivity, large color LCDs, directional keypads, and plenty of buttons. These features are precisely what is needed in controllers for programmable construction toys. Additionally, gaming handhelds seem like a great platform for visual programming languages that are commonly used for programming such toys. Gaming handhelds might even completely obviate the need for using PCs to interface with programmable construction toys and in doing so broaden the appeal of such toys.

Lego is aware of the potential of gaming handhelds as a controller for programmable construction toys. In fact, the aforementioned Media Lab paper specifically mentions the Gameboy as a promising candidate for hosting a portable programming interface. Lego is purposefully staying clear of the gaming handhelds. Lego's biggest problem with the gaming handhelds is that Lego will have to pay sizable royalties in order to deploy its software on the gaming handhelds. On top of that Lego will end up developing software for multiple platforms, and will also have to create separate product lines for each gaming handheld. This is too much trouble and because of this Lego wants nothing to do with the gaming handhelds.

This is where the gaming handheld vendors can win big. If one of the major players was to get into a joint-venture with FischerTechnik and introduce FischerTechnik's construction toys under its own brand, Lego will get completely cornered. Selling the toys will not be an issue for the gaming vendor as they will be backed by the brand name and the distribution muscle of the vendor. Moreover, as previously discussed, programmable construction toys sell best in video game aisles and the toys can go on video game shelves. Such toys can even be marketed as educational toys, and such a pitch will attract lots of video-game wary parents.

The gaming handheld vendor will win by default if Lego does not try to counter such a move, but even if Lego tries to counter the move the gaming handheld vendor will still win. Lego won't be able to boycott the platforms of any of the major handheld vendors, so it will be forced to introduce products for even its competitor's (the competitor is a major player by assumption) handheld gaming device. Lego's revenues exceed $1 billion a year and if 40-50 percent of that revenue was to shift to construction toys that require a gaming held, Lego will end up paying very sizable royalties to the very same gaming handheld vendor it is competing with.

Ultimately, the move to programmable construction toys is inevitable. The move is going to come with or without Lego. Sooner or later Lego's competitors will manage to bring such toys to the mass market, and if Lego continues to resist the move it will end up losing every thing it has tried to protect and more.

by Usman Latif  [May 03, 2006]

Thanks to Eric D. Burdo for reading drafts of this article.