4 ms·

> everything someone creates is always built upon the knowledge of others Right. Sure. These are the kinds of beliefs often verbalized by people who never risk

> everything someone creates is always built upon the knowledge of others Right. Sure. These are the kinds of beliefs often verbalized by people who never risked their own time and money to solve hard problems and had to face the risk and reality of the Chinese (or others) stealing the results of that effort. Here's the difference, the "R" in "R&D" is the most expensive part. The "D" is cheap. By stealing they get the "R" for free and can afford devastating predatory techniques to own the market. We devoted ten years and millions of dollars to solving a very hard problem. I decided we would keep all of it a trade secret. Why? Because, that was the only way to make "D" very expensive. Stealing something when the cost is just as high as having to develop it often scares away a lot of thieves. It's one thing if you are developing a new set of headphones that don't really have much in the way of technical innovation. It's quite another thing when you are solving a hard problem (which can be risky, very expensive and require a long and constant effort).

sgt101 11y ago
I think R cost is low, creating a poc is about 100th of the cost of a product (a proper product) but the risk of R is huge and therefore it's very hard to sustain and justify.
- pilom 11y ago
  While that is often true in software it is very seldom true in hardware. Think hard drives. The first HDD with HAMR (using lasers to heat up the platter to make it easier to isolate and flip a bit) required many millions of dollars to figure out how to do it. And when they started, no one knew if it was possible. Once they knew it was possible, creating a factory to build them was simple.
  - sgt101 11y ago
    Building a fab costs ~$8bn now, does Intel spend that in research? They report (like everyone) R&D, but I bet the majority is D. Delivering a product is very different from making a component (like a platter) - the safety testing, reliability engineering and all the other elements of development are expensive and labour intensive.
    - rebootthesystem 11y ago
      You are missing the point. You are thinking "expensive" is equal to "more money". Expensive, in this context, means far more than money. For example, risk. Sometimes company-ending risk. I said "R" is expensive because it is often packaged with uncertainty. You have no idea if something is going to work or if people will even be receptive to the product. Take SpaceX's landing rockets as an example. Hard and very expensive research. Lots of details nobody will ever know about. Lots of failures before success. And then, even with success, you have to make it repeatable and reliable. One the "R" is done, "D" or building and scaling is, in relative terms, cheap. Sure, it can cost billions to tilt-up a factory to build hard drives. Yet, you have a product and you have a market. You have an opportunity. You didn't have any of that before. Wright Brothers: Hard and expensive R&D requiring YEARS of dedication. Once everyone knew you could build such planes (and how to do it) the "D" was cheap because it was done. Sikorsky and helicopters is another AMAZING example of that. Helicopters got launched because of the Sikorsky's hard work and the unbelievable involvement of none other than Sergei Rachmaninoff (as in the pianist and composer) taking a leap and investing in the "R" that allowed Sikorsky to complete his work [0]. "R" is very expensive, in more than financial terms, that's why the Chinese copy rather than innovate. That we are allowing them to get away with it is, well, disturbing. [0] http://www.sikorskyarchives.com/Igor_Sikorsky_&_Sergei_Rachmaninoff.php http://www.sikorskyarchives.com/Igor_Sikorsky_&_Sergei_Rachm...
      - sgt101 11y ago
        My comment was "I think R cost is low, creating a poc is about 100th of the cost of a product (a proper product) but the risk of R is huge and therefore it's very hard to sustain and justify." Which I think you have provided some good supporting statements for. I must really be missing something here.. Not sure what though!
t0mk 11y ago
What is the hard problem you were trying to solve?
- rebootthesystem 11y ago
  I'll just say it was a very hard problem in imaging. A solution involving the confluence of hardware, optics, software and real-time FPGA-based very high resolution image processing. We even had to go as far as to develop manufacturing techniques that involved assembling electronics in a vacuum chamber. Given the application I can't say any more than that.

sgt101 11y ago

I think R cost is low, creating a poc is about 100th of the cost of a product (a proper product) but the risk of R is huge and therefore it's very hard to sustain and justify.

pilom 11y ago
While that is often true in software it is very seldom true in hardware. Think hard drives. The first HDD with HAMR (using lasers to heat up the platter to make it easier to isolate and flip a bit) required many millions of dollars to figure out how to do it. And when they started, no one knew if it was possible. Once they knew it was possible, creating a factory to build them was simple.
- sgt101 11y ago
  Building a fab costs ~$8bn now, does Intel spend that in research? They report (like everyone) R&D, but I bet the majority is D. Delivering a product is very different from making a component (like a platter) - the safety testing, reliability engineering and all the other elements of development are expensive and labour intensive.
  - rebootthesystem 11y ago
    You are missing the point. You are thinking "expensive" is equal to "more money". Expensive, in this context, means far more than money. For example, risk. Sometimes company-ending risk. I said "R" is expensive because it is often packaged with uncertainty. You have no idea if something is going to work or if people will even be receptive to the product. Take SpaceX's landing rockets as an example. Hard and very expensive research. Lots of details nobody will ever know about. Lots of failures before success. And then, even with success, you have to make it repeatable and reliable. One the "R" is done, "D" or building and scaling is, in relative terms, cheap. Sure, it can cost billions to tilt-up a factory to build hard drives. Yet, you have a product and you have a market. You have an opportunity. You didn't have any of that before. Wright Brothers: Hard and expensive R&D requiring YEARS of dedication. Once everyone knew you could build such planes (and how to do it) the "D" was cheap because it was done. Sikorsky and helicopters is another AMAZING example of that. Helicopters got launched because of the Sikorsky's hard work and the unbelievable involvement of none other than Sergei Rachmaninoff (as in the pianist and composer) taking a leap and investing in the "R" that allowed Sikorsky to complete his work [0]. "R" is very expensive, in more than financial terms, that's why the Chinese copy rather than innovate. That we are allowing them to get away with it is, well, disturbing. [0] http://www.sikorskyarchives.com/Igor_Sikorsky_&_Sergei_Rachmaninoff.php http://www.sikorskyarchives.com/Igor_Sikorsky_&_Sergei_Rachm...
    - sgt101 11y ago
      My comment was "I think R cost is low, creating a poc is about 100th of the cost of a product (a proper product) but the risk of R is huge and therefore it's very hard to sustain and justify." Which I think you have provided some good supporting statements for. I must really be missing something here.. Not sure what though!

t0mk 11y ago

What is the hard problem you were trying to solve?

rebootthesystem 11y ago
I'll just say it was a very hard problem in imaging. A solution involving the confluence of hardware, optics, software and real-time FPGA-based very high resolution image processing. We even had to go as far as to develop manufacturing techniques that involved assembling electronics in a vacuum chamber. Given the application I can't say any more than that.