All well and good, but we should also admit that under this criterion 95% of published math papers don't contain new ideas but are just filling in stuff that previous authors missed.