Computer Engineering: Adapting to the Future of Technology

[Forge]

We're hitting the silicon barrier so we might jump to a new type of computing, right?

So would my knowledge all go to waste if we move to optical computing for example?

 

[Joseph King]

I don't think it's ever a waste, necessarily. There may still be some things to discover in silicon based computers, but I do agree that the optical computing may be more lucrative, or maybe developing the quantum computer would be a good way to go if you have the dedication.

 

[AlephZero]

Whatever you study, some of your knowledge will get out of date or irrelevant if you work for say 30 years before you retire. So don't worry about it. Everybody else is in the same situation.

Of course if you have superpowers and you can really predict the future, that would be different

 

[SteamKing]

It's only stupid if you were planning to work in planar IC design exclusively. I think computer engineering involves more than that. While newer technologies might appear to be on the horizon, who is to say if these technologies will ever be mature enough to replace silicon?

 

[carlgrace]

It's only stupid if you were planning to work in planar IC design exclusively. I think computer engineering involves more than that. While newer technologies might appear to be on the horizon, who is to say if these technologies will ever be mature enough to replace silicon?

Planar IC design isn't going anywhere for the foreseeable future. Most 3D technologies are looking to connect planar ICs with through silicon vias or multi-wafer bonding. The skills will still be used.

Besides, computer engineering is more abstract that silicon engineering. The concepts are valid whether we are using silicon, light, bacteria, or an abacus for our computation medium.

Majoring in computer engineering isn't dumb at all. If you look at an introductory computer textbook from the 1960s you'll see that it is still mostly relevant today (some things are obsolete) even though it is 50 years old. The basics don't change much... the details do.

 

[Best Pokemon]

Whatever you study, some of your knowledge will get out of date or irrelevant if you work for say 30 years before you retire. So don't worry about it. Everybody else is in the same situation.

Of course if you have superpowers and you can really predict the future, that would be different

Math knowledge doesn't become out of date.

 

[carlgrace]

Math knowledge doesn't become out of date.

Sure it does. At the very least it can become irrelevant.

80 years ago a lot of engineering and math training involved hand calculating approximate solutions to differential equations. Not a very useful skill these days.

If you were an expert in that during the 60s and 70s when computers started taking over that role (and enabled much more accurate iterative solutions to diff eqs) you would have to update your skills otherwise *you* would become out of date.

Math is such a big field it is impossible to know all of it. The right tool for the job changes over time because the pressing problems change.

 

[Forge]

Thank you all.

 

[phyzguy]

Math knowledge doesn't become out of date.

Right, I still use my slide rule every day (NOT!)

 

[AlephZero]

Math knowledge doesn't become out of date.

Some of the math methods that I use every day at work hadn't even been invented when I was at university.

If you look at an introductory computer textbook from the 1960s you'll see that it is still mostly relevant today (some things are obsolete) even though it is 50 years old. The basics don't change much... the details do.

Hmm... computer science in the 1960s is going back to before classics like "Dijkstra, Hoare, and Dahl" (which shouldn't need the book title, any more than "K&R" does). But the principle of what you are saying is right, even if the timeline is a bit off.

 

[carlgrace]

Hmm... computer science in the 1960s is going back to before classics like "Dijkstra, Hoare, and Dahl" (which shouldn't need the book title, any more than "K&R" does). But the principle of what you are saying is right, even if the timeline is a bit off.

I was referring to computer architecture rather than software (based on the OP's question). Most of the earth-shattering developments in hardware design, for example the IAS Von Neumann computer, the Harvard architecture, Stretch (vector processing that led to the Cray), Superscalar architecture (CDC 6600) and the IBM 360 were built by the mid-60s. With the exception of RISC (and a few others of course) I think most of the big paradigm shifts in computer hardware had happened by 1970.

For software I agree with you. Interesting stuff, at any rate.

 

[Vanadium 50]

With the exception of RISC (and a few others of course) I think most of the big paradigm shifts in computer hardware had happened by 1970.

And RISC was 1975.

 

[Permanence]

Never a waste. You may not use everything you learned in school when you get into the work-force. Hell you may not remember much from school either. I think the whole point of a major is to get some exposure and demonstrate that you're capable of learning from a future employer.

Also they're always looking for people with experience in the 'old'. I know some companies still rely on old machines/old languages, and the freshly trained college graduate generally doesn't have exposure to that mess.

Now I've always felt that an EE major is always a better route than a CS major, but that doesn't mean a CS major is useless.

 

[TMFKAN64]

Very few computer engineers need to know the specific details of the underlying process that implements the circuit elements. For the most part, designing an optical computer would be very similar to designing a silicon one, until you got to the point of putting actual circuits together.

Now if we move to quantum computing, *that* would be more of a paradigm shift in how to design computers.

I wouldn't worry about the entire field becoming obsolete, although you certainly have to work to keep your skills up to date.

 

[Best Pokemon]

Sure it does. At the very least it can become irrelevant.

80 years ago a lot of engineering and math training involved hand calculating approximate solutions to differential equations. Not a very useful skill these days.

If you were an expert in that during the 60s and 70s when computers started taking over that role (and enabled much more accurate iterative solutions to diff eqs) you would have to update your skills otherwise *you* would become out of date.

Math is such a big field it is impossible to know all of it. The right tool for the job changes over time because the pressing problems change.

Right, I still use my slide rule every day (NOT!)

Using a calculator, slide rule, computer or paper and pencil shouldn't affect your knowledge of math.

Some of the math methods that I use every day at work hadn't even been invented when I was at university.

Can you give an example?

 

[AlephZero]

Can you give an example?

http://en.wikipedia.org/wiki/Daubechies_wavelet

 

[AnTiFreeze3]

Sure it does. At the very least it can become irrelevant.

80 years ago a lot of engineering and math training involved hand calculating approximate solutions to differential equations. Not a very useful skill these days.

If you were an expert in that during the 60s and 70s when computers started taking over that role (and enabled much more accurate iterative solutions to diff eqs) you would have to update your skills otherwise *you* would become out of date.

Math is such a big field it is impossible to know all of it. The right tool for the job changes over time because the pressing problems change.

He's talking about knowing math, not random techniques for solving math. Differential equations may have been solved by hand in the past, and that has changed, but the knowledge of differential equations is still very useful today. Similarly with logarithms, we used to have to find values of logarithms by hand, back when calculators weren't around, but logarithms have been and will continue to be important.

Math involves absolute truths; the truth can't be outdated.

 

[Turion]

Math involves absolute truths; the truth can't be outdated.

And when new truths are discovered, do people just know them without going to school?

Just because we might switch to optical computing doesn't mean that all my knowledge on silicon suddenly becomes wrong. It just means that I have to go to school to learn about optical computing just like a mathematician would have to go to school to learn new math.

 

[ModusPwnd]

Math involves absolute truths; the truth can't be outdated.

But the axioms that said "truths" are a function of can be changed or dropped all together.

 

[carlgrace]

He's talking about knowing math, not random techniques for solving math. Differential equations may have been solved by hand in the past, and that has changed, but the knowledge of differential equations is still very useful today. Similarly with logarithms, we used to have to find values of logarithms by hand, back when calculators weren't around, but logarithms have been and will continue to be important.

Math involves absolute truths; the truth can't be outdated.

Spare me the "absolute truth" stuff. The OP was interested if a degree in computer engineering would still be relevant if there are significant technological changes in the future. An engineering degree can only teach you so much, and there really isn't time to teach EVERYTHING. Of course differential equations are still useful, but the many, many hours spent in the 1980s learning tricks to solve them by hand is obsolete. A trick to solve an equation by hand is most certainly obsolete. These tricks are now out of date. QED. It would have been better to spend the time learning more fundamentals, but what is fundamental and what is a time-saving trick isn't always clear except in hindsight.

All in all, the vast majority of an engineering degree from a few decades ago is still germane. Some of it, though, is obsolete.

 

[AnTiFreeze3]

Spare me the "absolute truth" stuff. The OP was interested if a degree in computer engineering would still be relevant if there are significant technological changes in the future. An engineering degree can only teach you so much, and there really isn't time to teach EVERYTHING. Of course differential equations are still useful, but the many, many hours spent in the 1980s learning tricks to solve them by hand is obsolete. A trick to solve an equation by hand is most certainly obsolete. These tricks are now out of date. QED. It would have been better to spend the time learning more fundamentals, but what is fundamental and what is a time-saving trick isn't always clear except in hindsight.

All in all, the vast majority of an engineering degree from a few decades ago is still germane. Some of it, though, is obsolete.

My apologies for interpreting your words in the way that you wrote them.

Alephzero said that, no matter what you study, some of it will become obsolete and useless. Best Pokemon said that math knowledge doesn't become out of date. You said that it does; I responded to your claim that it does. I don't see the discrepancy that invalidates any of my response.

If you were to make the distinction that certain techniques to solve math problems become outdated, then you would be correct; however, that is the only evidence that has been supplied, so I fail to see how you are showing that math itself can become outdated.

 

[AnTiFreeze3]

But the axioms that said "truths" are a function of can be changed or dropped all together.

But then you're dealing with a different field of mathematics from the one in which the original axioms came from. Those truths still hold.

 

[ModusPwnd]

All truths hold with that distinction being made, absolute or not, mathematical or not.

Regardless, when it comes to employment (which is what this sub-forum is about) math can certainly become outdated. Its like the smart cow problem, employers don't actually need many people who are good at math. Its the useful and marketable techniques surrounding math that employers hire for. Which techniques are useful and marketable are in flux.

Generations ago being good at arithmetic alone was a marketable skill. Now that skill is obsolete, obsolete in terms of being a marketable skill.

 

[carlgrace]

My apologies for interpreting your words in the way that you wrote them.

Alephzero said that, no matter what you study, some of it will become obsolete and useless. Best Pokemon said that math knowledge doesn't become out of date. You said that it does; I responded to your claim that it does. I don't see the discrepancy that invalidates any of my response.

If you were to make the distinction that certain techniques to solve math problems become outdated, then you would be correct; however, that is the only evidence that has been supplied, so I fail to see how you are showing that math itself can become outdated.

Fair enough. I understand now that you don't consider methodologies for solving math problems to be "math knowledge". You've got a non-standard definition of "math knowledge", but that's fine. I think we're on the same page.

When I interview someone, the "math knowledge" I care about is whether the candidate can calculate the gain and phase margin of a feedback circuit, or can estimate the rms value of a signal from its peak-to-peak value, that type of thing. A deep appreciation of absolute truth doesn't help me meet my deadlines.

 

[TMFKAN64]

Math involves absolute truths; the truth can't be outdated.

True. But what changes is our notion of what constitutes *interesting* math.

Differential equations remain interesting. Methods of calculating approximate solutions by hand remain true, but are less interesting than they were before.

 

[AlephZero]

Those truths still hold.

There are no "truths" in mathematics. There are axioms, which are arbitrary, and made up by mathematicians. There are also rules of inference (i.e what is a correct logical argument), which are also arbitrary and made up by mathematicians. And that's all.

I think you are getting confused by the fact that a lot of math uses axions and rules of inference which correspond pretty well to the "real world" (whatever that is!).

The ancient greeks would probably have agreed with your Platonic philosophy of math, but things have moved on a bit since then, in both math and philosophy.

 

[AnTiFreeze3]

There are no "truths" in mathematics. There are axioms, which are arbitrary, and made up by mathematicians. There are also rules of inference (i.e what is a correct logical argument), which are also arbitrary and made up by mathematicians. And that's all.

I think you are getting confused by the fact that a lot of math uses axions and rules of inference which correspond pretty well to the "real world" (whatever that is!).

The ancient greeks would probably have agreed with your Platonic philosophy of math, but things have moved on a bit since then, in both math and philosophy.

But the point is that the statement obtained by the axioms and inference rules are true because they have been inferred. I have not said that the truths apply to the real world, simply that they are true in the domain of mathematics.

Regardless, any lack of correspondence between mathematical truths and truths in the real world does not mean that knowledge of math becomes useless. I do, however, have no doubt that some fields of mathematics are hopelessly lost and have very little interest or applicability.

 

[carlgrace]

But the point is that the statement obtained by the axioms and inference rules are true because they have been inferred. I have not said that the truths apply to the real world, simply that they are true in the domain of mathematics.

I think you're mixing up "true" with "self-consistent".

 

[AlephZero]

I think you're mixing up "true" with "self-consistent".

So do I. Let's take a simple historical example. When Euclid wrote down his axioms for geometry, it was clear (if you read Euclid's contemporary commentators, like Proclus) that these were considered "true" in an absolute sense. But even the greeks had problems with the parallel postulate, because it was much less obviously "true" than the other axioms.

From the ancient greeks onwards, mathematicians tried to prove or disprove the parallel postulate, by assuming it was false and trying to find a contradiction. Eventually, Gauss and his contemporaries showed this was a futile exercise, because if Euclidean geometry was self-consistent WITH the parallel postulate, it was also self-consistent WITHOUT it (the basic idea being to redefine a "plane" as "the surface of a sphere"). The assumption about self-consistency isn't really important, because adding another axiom to an inconsistent system doesn't make the situation any worse!

For a more modern example (starting in the 19th century), consider the following joke by mathematician Jerry Bona: "The Axiom of Choice is obviously true; the Well–Ordering Principle is obviously false; and who can tell about Zorn’s Lemma?"

If you don't understand the joke, do a bit of research - Wikipedia is your friend.

(Personally, I think the joke would have been even better if he had tipped his hat to Godel and changed "tell" to "decide").

 

[Carnivroar]

Colleges are not supposed to teach you job skills. It's supposed to teach you how to think and how to learn how to learn.

Corny, but most large IT/software coorporations don't care at all about the specific tools you studied in college - a smart person can learn anything.

If you're good enough to complete a CE degree then you're probably smart.