Computer Engineering and Computer Science (herein referred to as CE and CS) are two very closely related majors. To an outsider, the two are nearly indistinguishable and easily conflated, but they are actually fairly different disciplines. Hundreds of freshmen engineering students interested in one of these majors will naturally question themselves and others as to whether they have ‘picked the right path’. As somebody who chose the former over the latter and has ruminated over this topic (for perhaps too long!), I believe it is worth organizing and writing down my thoughts as a reference for others in the future at UIUC and at other universities. In a nutshell: computer engineers study the physics of computation and computer scientists study the mathematics of computation.
Part of the reason for the confusion is because the terms “engineering” and “science” are already synonymous in most minds. So let us instead look comparatively: first at the differences and then at the similarities. (If there are any left!)
Historically, Computer Engineering and Computer Science are identical in age because neither one could reasonably exist without the other. Despite others who may take poetic license and claim that civilizations with abaci are somehow evidence of computing’s ‘ancient history’, the first recognizable steps towards computing machinery were first taken by Charles Babbage with the invention of his Analytical Engine and Ada Lovelace who corresponded with Babbage and wrote the first algorithm for the machine. Right from the get go, we see the division of labor. Babbage, the engineer, creating the physical substrate for computation. Lovelace, the programmer, writing the code. It’s a bit too early to call Babbage the father of CE and Lovelace the mother of CS. But this leads to my first opinion: Babbage would have had a much easier time picking up programming than Lovelace would have had picking up engineering. Stated more generally: It is usually easier for a CompE to pickup a software job than for a CS major to pickup a hardware job. This not meant to be insulting towards Lovelace or towards computer scientists (or towards women for that matter). It is an uncontroversial fact in many other fields that the engineers who build machines generally will make good operators of those machines. An automotive engineer will most likely be able to drive a car with more finesse than the average driver and perhaps with ample track time could even drive at a competitive level, but even the most skilled NASCAR or F1 driver cannot reasonably design an engine, drivetrain, or gearbox without going back to school for specialized training. The same applies for aircraft: many Aerospace Engineers become pilots, but not every Blue Angel needs to have a PhD in Fluid Dynamics.
“So who cares? Why does this matter? Most computer scientists have no interest in hardware jobs.” Admittedly, this is really more of a useful point for the students leaning towards CE over CS. In plain English: if you want to learn hardware, do it now, or forever hold your peace. It is much harder to try and learn about Maxwell’s Equations, transistors, lithography, VLSI, and Architecture among many other things once you leave college. Many computer engineers end up doing purely software jobs for their entire professional career. Indeed, “software is eating the world,” as Marc Andreessen observes.
Faced with these odds, is it still worth pursuing a CE degree? I would say yes:
- Knowing how hardware interacts with software will allow you to optimize code for the machine, when algorithmic improvements start to taper off
- The core elements of Electrical and Computer Engineering: Signals, Systems, Control Theory and Information Theory are applicable in fields far and wide, from finance to biology.
- New computer architectures, paradigms, and models are easier to grasp when the underlying hardware is not a mystery. For example, GPU computing is an easier leap for computer engineers.
- A lot of software is written to simulate physics and to solve inverse problems. Such software is easier to understand and optimize when one has a background in electromagnetism, thermodynamics, quantum mechanics, etc.
- Computers are not always perfect, they can have bugs by design or due to noise. Many computer engineers are devoted to studying fault tolerant computing.
- Low power computing is increasingly becoming important with the advent of mobile devices. Understanding the impact of algorithms on power consumption is a classical computer engineering skill.
Of course, there are still the ‘pure’ CE careers, like computer architecture design and verification, but I hope the aforementioned points are hints as to why a purely software career path may still benefit from a CE curriculum.
That being said, CE is still not for everybody. There are many who really should choose the CS route. Before listing off the benefits of a computer science curriculum, one point should be made abundantly clear: computer science is not equivalent to software engineering. A classic ‘old school’ CS curriculum is not about learning how to program or how to write good software. Realistically, most CS sequences will have one or two classes on software style and engineering. But these classes are really to reassure potential employers that the students have actually touched a keyboard at some point.
As I said in the beginning, CS is about learning the math behind the programming; it is about algorithmic analysis, and the linguistics of describing a programming language that a machine can eventually understand after compilation. It is about using deductive logic to prove properties of algorithms and the languages that describe them, not about worrying how those algorithms impact transistors. Though, of course, most CS curricula will have a few courses on architecture. If architecture is ‘pure’ computer engineering, then compilers, cryptography, and algorithms are ‘pure’ computer science.
Now, that being said, most advertised “CS jobs” are not really pure, theoretical, computer science. Most are more accurately described as software engineering jobs. Which is fairly self-explanatory: the art creating software. I would say most big software companies like Facebook, Google, and Salesforce are looking to fill software engineering positions despite prerequisites that ask for a CS degree. A step below a software engineering job in prerequisite knowledge would be an Information Technology job, which usually requires on the job training as it is more of a craft than academic exercise. IT jobs historically have not required programming, but many are starting to require some coding experience as managing increasingly complex business IT starts to look like a software job of its own. I would say the main similarity between CS and CE is this: both can do software engineering and IT jobs.
So who would definitively benefit from a CS degree?
- Students who think Lisp is a good idea. (Sorry, I couldn’t resist.)
- Students that enjoy deductive logic and prefer math over physics
- Students looking to graduate in 4 years with very good job prospects
- Students that are interested in software engineering solutions rather than the computer itself
The two men that really embody the differences between CE and CS are John von Neumann and Alan Turing. I will spare repeating their full biographies here. But I believe von Neumann, a physicist, is truly the father of computer engineering and Turing, a mathematician, the father of computer science.
I think the content thus far establishes the philosophical differences between the two fields. There are, of course, practical and provincial differences depending on what university you will be enrolled in as a student:
- Your university may not offer both degrees (many smaller colleges can only offer CS)
- Your university may offer a degree in ECE, not in EE and CE separately (Carnegie Mellon University)
- Your CS department may have Computer Engineering as a specialization (Stanford)
- Your university may organize itself into a EECS department, where CE is somewhere in the middle (UC Berkeley)
- Your university may classify CE as a specialization in an EE degree
- Your university may be offering a Software Engineering degree but mislabeling it as CS (many community colleges are unfortunately doing this, hint: if none of the classes require proofs, it’s probably not worthy of being called CS)
Perhaps this post is really the result of historical accidents. The way departments did or did not divide CS and CE into two separate tents, or even if they had created tents to begin with. My advice is to take classes in the fields where you think the world is headed, and where you want to be. As a computer engineer, I am starting to shift my sights towards more biological applications, but I still feel I am in the right department especially for reason #2 that I gave above. Signals, Systems, Control Theory and Information Theory are really the key faculties of a good ECE curriculum that (I hope) will help me to solve problems that are interesting, fun, and existentially satisfying.