Read the title and skip the paper. :) Do you trust AlphaChip? The paper is a fun read -- very detailed and meticulous criticism of their critics. 

Claims:

- AlphaChip is very good at designing computer chips, even better than humans. 

- Igor Markov's criticism (in CACM) wasn't very fair because he didn't follow normal scientific practices. 

 "used substantially fewer compute resources (20x fewer RL experience collectors and half as many GPUs)" 

 "did not train to convergence (standard practice in machine learning)"

"and evaluated on test cases that are not representative of modern chips"

"This would be like evaluating a version of AlphaGo that had never seen a game of Go before (instead of being pre-trained on millions of games), and then concluding that AlphaGo is not very good at Go." 

Interesting that they wrote an arxiv paper to defend a Nature paper. Now how do we peer-review this paper?