Slashdot Mirror
How They Make LEGO Bricks
harajukboy writes "Businessweek.com shows us how the famous LEGO bricks are made. Among the new facts I picked up was that LEGO is the largest tire manufacturer in the world, and that the process is so air tight that only 18 of 1 million pieces are considered defective." I knew I was getting old when I first realized that these kids today with their modern legos have it too easy, what with all those crazy custom pieces. Why, when I was a kid, we had to use our imagination to build stuff.
Didn't LEGO outsource their fabrication off to some other company earlier this year? I'm pretty sure it was a fairly long transition, so we may not be seeing these new pieces yet, but the sets I've bought in the past year seem "different" somehow. Colors don't seem as solid as they were years ago, and the plastic feels softer. They still snap together pretty well, but they don't seem to fit against each other as well and seams can be much bigger than I remember.
While I'm sure the machinery and manufacture process isn't changing, it would have been nice if the article could have commented on the changes being made in response to the restructuring LEGO has been doing the past couple years. It's pretty obvious to me that things are changing, but it'd be nice to have it documented.
Exactly. I'm no export on Six Sigma, but my company is heavily into it, and I had to go to the introductory course. Opertunities would include colour, shape, hardness of the plastic, just to name a few. One perfect lego block could represent a dozen or more opertunities for error.
Lo and behold the slideshow link gives you a 404 File Not Found.
I was able to view the slideshow. Nothing impressive, but nice pics.
this presentation is better, including short video clips of the machines in action. It's also more fun. Requires Flash, though.
same presentation without it opening in a popup window
Bureaucracy.
At least in GE's implementation of Six Sigma. They found a way to take what is essentially the engineering version of the scientific process, wrap it in so much red tape that it is unworkable (a 12-step process that really had 15 steps) , and put it in the hands of every worker in the company. Originally they gave bonuses for doing it, but eventually they took those away and declared "Thou shalt not get a raise without a Six Sigma Project." What ended up happening is that people refused to make any process or product improvements unless they were part of somebody's (preferably their own) Six Sigma project.
It was ridiculous. You ended up with one person optimizing a part of a process, while the person in the next cubicle was eliminating the entire process in favor of a more unwieldy one. Then, six months later, somebody else would start a new project that essentially put the original process back in place. Of course the problem was that they were using a distinctly product-oriented procedure, and trying to use it to solve process problems.
Don't even get me started on the math. They would assume normal distributions for everything. Never mind that one of the steps was to prove normalcy. If that test proved it wasn't normal, you were instructed by your "Black Belt" to assume normalcy anyway -- even if a Weibull distribution was clearly the correct choice (like in timed exercises). Idiots, I say. And then they had PHB's (called "Black Belts" and "Master Black Belts") trying to tell engineers how to do math, when they didn't even know how to use a simple Q test. If they saw a data point that didn't support their theory, they just called it an outlier, and deleted it.
You'd think after nearly two years of not working at GE, I wouldn't get so wound up about it. I guess as an engineer, it really gets my goat when people use math improperly.
Hmm interesting that it's almost the same in degrees Fahrenheit and kelvin. That inspired me to figure out that:
574.5875 degrees Fahrenheit is exactly 574.5875 kelvin.
I'm sure Intel or AMD would love to hire you.
Or probably not. They have much lower yield rates because their processes are constantly shrinking, often to double the number transistors in a given area every year and a half. I read there's an adage in the DRAM industry that too high of a yield is bad because it means there's capacity potential not being properly exploited. If they didn't keep pushing much faster and much higher capacity products, I think they could do six sigma.
As others have noted, six-sigma has been a failure in business. You get reduced defect rates, but the cost in getting such a low defect rate is generally so exorbitantly expensive that you are better off recycling the rejects than spending the money to eliminate rejects. There's a similar joke about ISO90001 that you get half the defect rate because it cuts your productivity in half, or practically doubles the cost per part because of the beauracracy involved.