I don't think this an adequate analogy. SVMs never completely captured the field. There were also random forest, boosting and other promising algorithms that de

I don't think this an adequate analogy. SVMs never completely captured the field. There were also random forest, boosting and other promising algorithms that delivered results "in parallel" (within the same time-frame). Also, the attitude that ANNs are "all we need" was there even before deep learning delivered all the recent SOTA results. I remember Patrick Winston commenting on that in one of his lectures in 2010(!).