Yes, incremental novelty of the sort where you go somewhat further down an apriori promising path and show some somewhat stronger results than the previous paper.

But I don't think any community ever has truly rewarded dissent. It's just that sometimes a truly valuable idea will push through in spite of the resistance to it. Case in point - around 2014 I knew CS/AI/ML profs who were very dismissive regarding deep learning and thought it to be a dead end and basically missed the boat and took them very long to get up to speed again. A few years before that, and it was difficult to get papers published that used neural nets. Everyone knew that the modern way was the theoretically well-founded kernel methods and similar techniques.

----

There's no overarching one thing called science. There have been over the last 150 years a few towering results that propelled our understanding of the universe and our technological capabilities forward immensely, on whose wind we are coasting today -- but many of them came before the current professionalized, job-ified massive system we know as "academia" today. Without peer review, in informal letters between gentlemen scientists persistently pursuing topics for their leisure, often with extreme concepts of "work-life-balance". We have no idea what exactly brought about those successes, and so we are building a cargo cult, somehow imitating it, LARPing it. Thousands of papers, salami-slice publishing, citation metric-chasing, quantification and metrification, incentives to hop from place to place, endless grant writing and documentation/administration. This is a very specific system that was mostly created and solidified over the last 50 years. It should not be equated with science itself, which was already a very productive endeavor hundreds of years ago, while being intimately intertwined with mysicism, alchemy, esoteria and theology. This particular utilitarian paper factory of today isn't equal to "science", even if it has taken over the buildings.