In my opinion, you need to understand the difference between hard work and smart work. Software development is like an art.
A person may spend 18 hours a day to create something that just works while another may spend just 8 creative hours to create a masterpiece.
There are many disadvantages of being a hard worker / workaholic:
* The expectation of your manager will get elevated and if one fails to be consistent in working for long hours, it's a negative.
* One does not have much room to accommodate something else at times of need. Think of an outage, you have already worked for 18 hours and fail to deliver at that crucial time.
* It's very difficult to innovate in an insomniac state.
Remember, doing hard work is easier that smart work. It requires much more learning and thinking. But, if you are able to do it, you will be able to contribute much more.
Appraisal based on relative performance: This sounds logical. If a person contributes better to the company's goal, he/she should be appreciated for that. Think yourself being that person it should make sense.
Note: I am not taking the political aspects of your office into consideration. If your manager loves late night availability, long emails at weird hours, it is his weakness. Probably, you can be smart and automate sending 'corporate bullshit' emails at night ;)
>Appraisal based on relative performance: This sounds logical.
I don't think it does. People should be rewarded based on the absolute value they bring to the company, not on their relative ranking.
For example, relative ranking means that you always want to work with the worst possible teams (so you are relatively the best) not the best. It also means that politics becomes more important than ability.
I agree that total value is a good ideal, but it's really hard to measure in some cases. Forgetting the inaccuracy of things like LoC, # of bugs fixed, etc. How do you measure the performance of someone who isn't directly on the product value chain? graphic designer delivered, office manager, technical writers, etc. Sometimes the most quantifiable and accurate method is by comparison.
