We need to replace the math courses in a CS degree with something more useful. Maybe some classes on Javascript, NoSQL, and whiteboard coding.
We need to replace the math courses in a CS degree with something more useful. Maybe some classes on Javascript, NoSQL, and whiteboard coding.
I'm not serious. The primary thrust hat people who have taken the math required by an undergraduate computer science or engineering degree can probably see at a glance that the typical startup salary offer is shit. Thus, I sarcastically suggested we should remove math from the CS curriculum in order to make graduates less able to understand this.[1]
I did so in a way that was also supposed to caricature the "programming doesn't need math" crowd, but apparently I came off as a member of the crowd. I was hoping that my sarcasm would be indicated by throwing in "whiteboard coding" as something a degree should cover. That so many people thought what I intended to be caricature was serious worries me.
[1] Hence the reason I said "abundance of math skills"; potential candidates know too much to be suckered.
For many (including myself) what we do is nearer to been builders than civil engineers and beyond a basic grounding time would be better spent on the other skills that make up a good developer.
On the other hand, these companies do need talent who never the less have skills that are rleatively rare. They shouldn't need to pay engineers the $200k+ salaries of big companies, but expecting to get median quality engineering without paying more than $110k/yr is as ridiculous as a non-degreed "webdev" expecting $180k/yr because he can slap some customizations on bootstrap and poke around a Ruby codebase.
The last piece is in many start-ups I've seen the programming team will own or co-own Analytics because of a lack of statistics skills on the business team. Again with a small team not having a solid math and stats background has huge implications on the data analysis portion of roles.
Aside from that, what the heck would you teach in a course on NoSQL? The theoretical underpinnings of key/value mapping? Proper denormalization of data? I'm honestly curious here.