There is an interesting tension in software between measurement and validation. The two concerns are often conflated, resulting in software systems that cannot

There is an interesting tension in software between measurement and validation. The two concerns are often conflated, resulting in software systems that cannot measure the world as-it-is. The world (more precisely, measurements of the world) may be in (and often is in) an 'invalid' or 'inconsistent' state. Software can only help you detect and resolve such states if it is capable of representing them.