Look, I'll be intellectually honest and give reference to that link. Snowflake would not be where they are if they didn't innovate. I was and I am still a big Snowflake follower, but I can't help but see them follow the path of Oracle, Terradata, and Redshift when it comes to performance. In my days of adminestring Redshift I had similar complaints, which is why we moved over to Snowflake.

Nontheless, here is why adding and talking in depth about cross-platform comparisons is not something I can do:

1. I can't verify Snowflake's claims or make a ocmment on them. None of those features seem to be something that would fix the more general types of query patterns I see. For example, the fix to substring searches is really only something that applies to searching large corpora and the compilation time was never really an issue for us as consumers. Compile type for large queries is usually less than 5%. And for very large is less than 1%.

2. I can't make cross platform comparisons without dedicating time and resources. Last time I ran a benchmark against Snowflake and Redshift with TPC-DS I had to pay several hundred dollars. You can't reasonably make a claim to a person on the internet to put in so much sacrifice and effort.

3. Complains and critique go a long way in making the company change and innovate. Multiple people I know complained about the "import CTE" issue and the worklload manager, yet nothing is being done. If you don't see the value in my article, don't read it.

Nontheless, if people start discussing some performance issues in public settings instead of private, I expect Snowflake to feel more responsive and hopefully in the next keynote they can announce a 40% gain in performance.