I feel like this article’s advice is going to fall on deaf ears. Those that have been through all of this before and know to use boring things like MySQL and Rails/PHP/Java are already going to do it, and probably could have written the same article telling this guy to do it when he was developing his product. Some people don’t appreciate stability and maturity in a tool, they see those tools as outdated and their use case is so radically different than everyone else’s that they must use something new (because surely, no one has ever created a product/startup as innovative as mine!). They just want whatever is shiny and new.
One thing I’ve learned from being around the web for a while is that everything that’s old is new again. There are an endless stream of newbies that would love to learn all these lessons over again.
A few of my blog posts that are just regurgitating the Ruby docs got a few thousand views, and a few ‘wow, I never knew that!’ emails.
This is basically how it works in computer science. When multiprocessor research was in full swing and they realized memory consistency models were flimsy and locking mechanisms were extremely necessary, the database researchers laughed and laughed. Because now the multiprocessor people were writing papers and “inventing” algorithms the database people invented decades prior for data and lock management, but the new guys had no idea. They were excited to be able to add something “new” to the architecture/os discourse. The prospect of doing something new continues to cloud our judgment over doing the dry, boring research to find out what is already there and what our choices are.