> My career is just a big middle finger pointed at this dichotomy.
I only saw your comment after the parent it replied to was flagged/removed. This section of your reply jumped out to me, care to elaborate?
My experience (undergrad, minimal research -> industry) has shown me that the gap is quite real, but both sides are important. They simply align to different goals. The most extreme trope of this in tech is industry engineers making fun of researchers who produce systems which "only work on paper" (paraphrased, some meaning is lost in that phrasing).
I find that criticism too harsh (and rude), but it does highlight the differing goals: industry frequently needs to build systems which 1) work, and 2) scale, among other things. Academia can afford to push the cutting edge by compromising on constraints real world systems have to obey. It's the classic exploration vs. exploitation problem. Industry thinks the exploration isn't important because they need to optimize the local maxima/minima now. Academia cares about the global maxima/minima, pushing the state of the art, exploring new solutions in the hope of advancing knowledge and understanding.
In the largest companies, the cost/benefit analysis shifts back toward exploration. Google, Microsoft, Xerox PARC, etc. And the natural extension of academics elsewhere pushing the state of the art is finding new optimums and bringing that back to industry (e.g., as a startup). So I feel there gap is real, with lots of symbiotic back and forth.
I don't 100% agree, but let's play with "academia is exploration, industry is exploitation".
How would you ensure you explore something useful? How would you ensure to transfer what you explored, so you can exploit that knowledge? What if you exploitation is stuck in a local minimum, how would you ensure you send feedback back to exploitation?
I feel a more useful model is Technological Readiness Level (TRLs). Academia deals with lower TRLs, while industry deals with higher TRLs. Somewhere in between there is -- there needs to be -- a huge overlap to ensure a proper continuum.
This can also be phrased in a risk perspective. Lower TRLs are extremely high risk, almost like throwing money out the window, but have extremely high societal reward. Think curing cancer or terraforming Mars. These are best funded via grants. High TRLs are low risk, low rewards; issue some bonds. Somewhere in the middle is VC territory.
But how? PhD students doing industrial internships; PIs doing industrial sabaticals; engineering managers becoming adjunct lecturers or giving guest lectures; academia and industry working on common projects; making patents could as much as papers on an academic CV.
Is anyone in the world either exploring or exploiting? Then why don't we create a system that allows people to quickly switch hats, as required to achieve larger goals?
I like your comparison to TRLs. I think we’re saying the same thing, actually! I don’t mean that one is exploitation and one is exploration exclusively. Re-reading my post, the phrase “gap” might be a misnomer — it’s more like a balancing act of priorities.
> How would you ensure you explore something useful? How would you ensure to transfer what you explored, so you can exploit that knowledge? What if you exploitation is stuck in a local minimum, how would you ensure you send feedback back to exploitation?
I’d argue you can’t necessarily be sure. Not all research works out. But we need to ask questions and get answers to move forward, even if what we discover may not be worth exploiting (getting tired of my own phrasing). Getting stuck in local min/maxima is part of why we need both angles. This is the flip side of what I said about industry needing to build systems that work — lower budget for risk tolerance, but also increased likelihood of getting stuck in a local optimum.
It sounds like I’m using those terms to describe the same spectrum you are with TRLs. At least, I wholeheartedly agree with everything you said — the “but how” portion sounds exactly like the symbiotic overlap/crossover I had in mind.
I guess to answer my own original question, it sounds like you’ve swung back and forth a few times?
> Then why don't we create a system that allows people to quickly switch hats, as required to achieve larger goals?
Preaching to the choir ;) and appreciate your perspective
> I guess to answer my own original question, it sounds like you’ve swung back and forth a few times?
Indeed, I couldn't decide if I want industry or academia. What I (think I) want is to experience the whole innovation pipeline, from TRL 1 to 9. I'm now working 80% in industry and 20% in academia, and it satisfies my intelectual needs, both to explore moonshots and to get sh*t done. :)
I only saw your comment after the parent it replied to was flagged/removed. This section of your reply jumped out to me, care to elaborate?
My experience (undergrad, minimal research -> industry) has shown me that the gap is quite real, but both sides are important. They simply align to different goals. The most extreme trope of this in tech is industry engineers making fun of researchers who produce systems which "only work on paper" (paraphrased, some meaning is lost in that phrasing).
I find that criticism too harsh (and rude), but it does highlight the differing goals: industry frequently needs to build systems which 1) work, and 2) scale, among other things. Academia can afford to push the cutting edge by compromising on constraints real world systems have to obey. It's the classic exploration vs. exploitation problem. Industry thinks the exploration isn't important because they need to optimize the local maxima/minima now. Academia cares about the global maxima/minima, pushing the state of the art, exploring new solutions in the hope of advancing knowledge and understanding.
In the largest companies, the cost/benefit analysis shifts back toward exploration. Google, Microsoft, Xerox PARC, etc. And the natural extension of academics elsewhere pushing the state of the art is finding new optimums and bringing that back to industry (e.g., as a startup). So I feel there gap is real, with lots of symbiotic back and forth.
Broad generalizations, but not unfair I think.