Completely agree, sure you can make superficial progress, but then you will need to further improve, and without deep knowledge of the underlying mechanics you will be lost.
Also rather important is to know what paths not to pursue in the search for a solution. Basically a solid understand of bias/variance will carry you far and help you avoid doing a week of work for an optimization that doesn't actually pan out.
Are you talking specifically about research? Because in my experience the time and $ delta needed to bring a classification model from say 93-96% accuracy is not worth it to most businesses. So all your special "deep knowledge" is irrelevant in most use cases
If you're in research I would assume a deeper background or at least an environment where you are encouraged to develop your background. In industry, yes, time to value is important, but it seems clear that in some part of the market competition will require squeezing out those remaining % points of performance -- a company scaling up with poorer accuracy would have more problems...right?
Do you have experience in industry? I would like to hear more from your perspective?
>In industry, yes, time to value is important, but it seems clear that in some part of the market competition will require squeezing out those remaining % points of performance -- a company scaling up with poorer accuracy would have more problems...right?
But I wasn't talking about a poor model vs a good model. Like I said, going from 93-96% accuracy is generally not going to have a lot of value add.
>Do you have experience in industry? I would like to hear more from your perspective?
Yes and in my experience if your model has lift over random and has a positive roi vs doing nothing it's usually going to be worth building and implementing it. In the fields I've worked in the models are not in direct competition with models from other companies so if Co. A's model for some task is getting 95% accuracy and Co. B's is getting 91% there's not much cause for concern. If Co. B's model is still generating lift over a random guesser it's worth having in production.
For most consumers I doubt they could even tell the difference between service like that. If Walmart's product recommendation is 2% better than Amazon's I find it extremely unlikely most consumers could even tell and Amazon's primary concern is whether the model is driving increased sales and not whether it's stealing customers from Walmart(the model, not generally).