The article helpfully considers the issue of logic being duplicated across layers, and questions the choice of potentially unnecessary layers of business logic

The article helpfully considers the issue of logic being duplicated across layers, and questions the choice of potentially unnecessary layers of business logic being in the architecture at all. I wonder, though, how this changes in a modern world of web services like Amazon's, where your web host provides REST APIs to talk to your database (e.g. DynamoDB), and you can serve your SPA as static files out of an S3 bucket, not needing to worry about server-side code at all. Taking out the server layer altogether may not work if you need a specific user auth model which your web services provider doesn't support, but this simplified serverless model has the advantage that attackers can't hack your server code if you don't have any server code.