Advertisement

Blog

Preserving the Holy Grail of Institutional Intelligence

The first generation of analog engineers started retiring in the mid-1990s, about 30 years after Bob Widlar of Fairchild Semiconductor designed the first widely used analog IC (the μA702 op-amp). Dave Talbert, a process engineer, was a key player back in a time when there was a symbiotic relationship between process and design. Those were the days.

Since the μA702 made its debut, an enormous amount of analog IC knowledge has been created, forming the backbone of the multibillion-dollar industry we have today. By the way, I'm not just talking about analog IC design; I'm also talking about system-level and subsystem-level design. Obviously, a lot of that knowledge has been saved and passed along to generation after generation of design engineers.

I've recently been wondering about how much of that knowledge (the secret sauce part) exists only in the gray cells of graying EEs — and whether it will be lost. Would it even make any difference if it were lost?

From what I've heard rather consistently over the past 20 years, it's getting easier to be a successful design engineer. A major player in the “recipe-fication” of system design is integration. Creativity is still highly valued, but in software, which, oddly enough, is based on digital technology.

My argument, which I will admit is pretty general, is that technology trends eventually reach a point of diminishing returns. The effectiveness of software configuration for increasingly massive chips will eventually be tapped out. At that point, we will see a new blossoming of hardware design. If my theory is correct, it would be a good idea to capture all that secret-sauce knowledge.

My first internal counterargument would be “Well, all that knowledge has already been captured in publications such as the Proceedings of the IEEE, hasn't it?” I think not. Competitiveness in the electronics industry has created thick, high walls to information sharing. These walls have been codified into legal documents that make it very clear who owns the intellectual property that springs from the eternally fertile minds of engineers.

There seems to be a bit of frustration about this turn of events. It is, for example, a motivating force behind the open-source movement. I should add that there are good reasons for companies to protect intellectual property, and I am not opposed to that. On the other hand, when lawyers become involved, we tend to move into a zero-sum game. The right level of sharing can benefit everyone.

In short, what I have in mind is a web-based, multimedia knowledge base. We already have that to an increasing degree on YouTube, but there is a strong bias toward current stuff. I'm talking about knowledge that may have been created decades ago but remains eminently useful — and may be even more useful in the future. Storing this comprehensively would require money.

Now, if I were a VC, I might be thinking to myself: “These guys made me rich many times over. Maybe it's time to do something for them.” Will that happen?

15 comments on “Preserving the Holy Grail of Institutional Intelligence

  1. jkvasan
    April 18, 2013

    @Jack,

    “The effectiveness of software configuration for increasingly massive chips will eventually be tapped out. At that point, we will see a new blossoming of hardware design. If my theory is correct, it would be a good idea to capture all that secret-sauce knowledge.”

    The secret-sauce would have to definitely come out breaking the barriers of intellectual property rights. This can happen only when the concerned magnanimously release this 'sauce' for open access. Would this fall in the ears of those who matter?

  2. Jack Shandle
    April 18, 2013

    Intellectual property rights have good consequences (they make it worth someone's while to invent new things) but they can also stifle creativity when they make it difficult to leverage the original idea. I think it is important to remember that intellectual property rights are not absolute. They are typically subject to time limitations, for example, and some cannot be exercised when the IP is used for educational purposes (e.g. photocopying pages of a text is OK for academia). I imagine there must be some case law about intellectual property that resides exclusively in an engineer's brain – which is the sort of IP I'm talking about in my blog. Typically, IP has a legal instantiation — a patent or copyright. I don't see how an engineer who makes a video that imparts some tricks of the trade is necessarily violating a patent or copyright unless the particular trick being demonstrated has somehow been mentioned in a patent.

  3. Jack Shandle
    April 18, 2013

    Scott–

    Thanks for your post. I agree with all of your points. The debate about intellectual property has been going on for a long time. The primary reason for this is that the issue has never been properly resolved.

    It's a shame that the law isn't as innovative as technology. I agree that journal articles should be not be subject to strict copyright protection.  As you note, a lot of the research reported in journals is paid for by tax dollars. But there is a reasonable case for giving the publisher an economic incentive to publish the journal. One option would be to provide a financial incentive through subsidies or some other means. Another option would be to amend the length the copyright is in force and/or make other changes in definition of a copyright. When I said earlier that the law is not very innovative, I was thinking that more sophisticated definitions of copyrights might well serve the interests of all parties.

    I'm reminded of the almost comical – but virulent – opposition of the music industry to dissemination of songs over the web a decade ago. As things turned out – and people were predicting this at the time – more money can be made by selling each song at $1 (with zero distribution costs) than by manufacturing 100,000 or more CDs containing 20 songs, packaging them, distributing them and handling the returns etc.

     

  4. eafpres
    April 18, 2013

    Patents are public. The problem is, a lot of important details aren't in the patents. Big companies develop patent portfolios to protect right to operate, for defensive reasons, and to block other companies. They don't do it to teach techniques. 

    A lot of important material is never published or patented. So-called trade secrets are at great risk from retiring experts, or even turnover. I've seen cases of process transfers taking months to forever because the key persons are no longer around. 

    Videos are great but a lousy way to document high tech details in my view. When was the last time you listened to a talk at a meeting and walked out confident you could duplicate what was reviewed? Those are 45 to 60 min in-person talks. Now try it with You Tube. Open courseware platforms like Coursera are probably closer to what is needed, but the cost is much higher to generate that quality. 

    So putting it all together seems like an opportunity to offer a good platform for “knowledge archiving and retrieval or training” but I would not expect companies to pay the bill for anything other than systems for their own use. 

  5. eafpres
    April 18, 2013

    I am as frustrated as anybody with the journal costs.  As a small consultancy, I simply cannot afford to subscribe to all the journals in which material of interest appears, nor can I afford the very high per-copy download fees.  If they would go to something like $5 I would probably spend a few hundred $$ a year.  But I won't pay $35 for something just to get the full paper unless it is really critical to a specific project.

  6. Jack Shandle
    April 18, 2013

    It is odd how things work out. The people who need information the most tend to be priced out of the market. I confess to not knowing much about the pricing model used by most journals. My guess is that in many cases it has not changed in quite a few years. Two decades ago, the primary purchasers were institutions — academic and corporate. Under that scenario, the “pain point” that determines a price point would be fairly high — too high for an individual. I believe that many professional societies (IEEE) give a discounted rate to members — but apparently it still doesn't help very much if a consulting engineer, for example, would like to have numerous papers from numerous societies.

    My first thought for a solution is to create a buyers' consortia. That is, a bunch of interested parties get pool resources for a single subscription to multiple journals. I don't know if this has been tried — or that such entities exist, or if it's legal — so if anybody can provide some information, I'd appreciate it.

     

  7. Netcrawl
    April 18, 2013

    @eafpres I think the figures makes sense, I'm sure there always  some good reasons for a certain company or organizations to protect their works. Its reliable and effective, its better to spend in those particular stuff especially if you chasing something and those things could help you.   

  8. sonic012
    April 18, 2013

    That's just dumb. How else could I put it?

    First,, I will be technical. The rate at which you can process a signal will never be faster in digital than in analog. It will never happen. What is happening is the analog to digital processing rate is going up. But even that requires the experties of an analog engineer. 

    Digital and software engineering is a joke by comparison. You don't even need a brain to do that stuff. 

    As IP goes. I guess I'm on the fence on that. I have been at companies where they didn't want to patent something, because it gives the possibility of giving away information. So we didn't even patent it. It's a business risk, to patent or not. Either way it is a risk. 

    Regards,

    alex

  9. Jack Shandle
    April 18, 2013

    Granted that today You Tube is more appropriate for sharing some cute trick your cat does — but I would also like to point out that a video is a video is a video. In other words, have we seen the end of innovation at You Tube?  I doubt it.

    I checked out Coursera and even after a fairly cursory look it seems to be clearly a better vehicle for saving the analog know-how that resides in the gray cells of engineers who are now retired or soon will be. Coursera seems to have a problem with its funding model, however, and that may be more of a positive if the two ideas have synergy. Even though there are only a few of us are participating in this thread, there seems to be a consensus that the analog knowledge we are in danger of losing is pretty valuable stuff. Wonder what Coursera thinks about this.

     

  10. Jack Shandle
    April 18, 2013

    Alex,

    Thanks for your post. Although I would not characterize digital and software engineering as “a joke,” I agree that analog is greatly underappreciated and that some of that analog knowledge that is in danger of slipping away should be saved (that was more or less the point of my blog).

    There will always be debate about IP protection because there is no clear winner of the argument. I'm tempted to make a comparison to composers of music — classical music, in particular — who have been happily borrowing from each other for 300+ years. Instead of suing for violation of copyright, they call it “homage.”

    Is there something we can learn from this perspective other than citing the “starving artist” cliché?

     

  11. sonic012
    April 19, 2013

    Hi Jack,

    So I've done both, analog and digital engineering extensivley. And it takes one who has done it to know the difference… C++ VHDL.  Digital logic. Versus mixed signal circuits, or RF circuits operating at 20GHz. It's not the same.

    Now on to your point about business. I don't think that knowledge is being lost. Yes a company has to decide when to patent something and when not to. It's always a business descion that you have to make, And I don't believe the free market is failing us. 

    Regards,

    Alex

  12. Jack Shandle
    April 19, 2013

    I don't think the free market is failing us either but I'm also more concerned about knowledge that has not been patented and that it should be — and can be — preserved.

     

    Although this is certainly not a perfect analogy, perhaps we can learn something from the legal profession, which has a repository of legal knowledge contained in its case law. As far as I know, no one particularly owns “case law” but it is widely available (free if you go to the local university law library) and the legal profession could not function without it. Engineering has journals and text books and other sources but all this seems to fall short of what the profession could create if it only had the right vehicle. In particular, I think there should be some way for knowledge that might be lost to be saved.

     

  13. sonic012
    April 19, 2013

    Well Jack,

    You got me on that one about the musical industry divulging information. That was hilarious. 

    I just don't agree that information is being lost. It's almost the opposite. The problem is that people don't seem to have the desire, persevernce, or intelligence to learn. I don't think it's about IP.

    Of course a company patents something with the least detail to give away what they are actually doing.. While giving the most detail so they can defend it if they realize their invention is being infringed upon. There's nothing wrong with that.

    Your plight is a storm in a tea cup.

    Alex

  14. jkvasan
    April 23, 2013

    “First,, I will be technical. The rate at which you can process a signal will never be faster in digital than in analog.”

    I find it difficult to accept. Digital Filters implemented through code can be faster than standard analog filters. Component related inaccuracies do not come in picture when a digital filter is implemented through a microcontroller.

  15. Brad Albing
    April 30, 2013

    Sometimes of course the techniques are specifically not patented because they are trade secrets (think formula to Coca-Cola). I don't see any way around that issue. If a company chooses to keep its secret sauce secret, what can you do? Except maybe reverse engineer it. That might be OK – since it's not patented.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.