I once read and author who claimed that goldbricking, malingering and sabotage were really just market effects. When employers pay too little or working conditions are too harsh, employees naturally find ways to remedy the problem. There seems to be a similar mechanism, here.
The pinnacle of this has to be the competitive incentive programs used at certain large software firms. In a competitive review system, the best policy is to passively sabotage your coworkers. This is robust even in the case where you are the rare superstar who has all the knowledge (in which case you simply hoard your knowledge). Those that try to be stars simply attract more coworker subversion. The countervailing factor is the general misery caused by this system. Employees will continue to sabotage until the work environment is as bad as they are willing to tolerate at the prevailing pay level. Thus, the only guaranteed result of a competitive review system seems to be maximally low morale and the majority of employee effort spent on gaming the system.
The late W. Edwards Deming (one of my heroes) proposed that the solution to this (in a manufacturing context) was to have a collective reward system (e.g., profit sharing), and to treat high and low performers as special cases. High performers should be examined to see if their techniques can be generalized across the organization, and low performers are examined to see if their performance is a result of the system or (rarely) a cause idiosyncratic to the low performer.
Many of the software types I've spoken to balk at such a method being applied to their organizations, because "creating software is not a production line process." This seems specious to me.
Have any of you if you've worked in a software house that had a reward system in place like the one Deming advocates?
The problem with enforcement is, it doesn't scale. In Joel's case, Joel and Michael are the ones monitoring the sales department (probably not more than a few guys) and Joel's interests are perfectly aligned with the company, so that's fine. But the second there's a layer of management in between, the Sales Manager (who is also getting paid on commission)has an incentive to let the sales people cheat as much as he is capable of hiding from the higher-ups. So I guess the approach is: use incentives, but only if the sales team has to report directly to people who own equity in the company?
I mean, I guess you could set up a system where you track each sales person's stats to make sure that their customer return vs exit ratio (or some other metric) doesn't fall below company average, and maybe that's what you have to do, but all that is is a better approximation of proper incentives. How much overhead is necessary until you have it right, and how much is that overhead going to cost you?
Amazing. There are three anecdotes that form this article. In all three the "employees will game incentive programs" element is pure speculation by the author.
Obviously he is paid a commission by the anecdote and thus makes them up thereby subverting his employers best laid plan.
How about paying employees directly for debugging the incentive plans: Pay them for reporting how such a plan could be subverted, and then adjust the plan accordingly.
I enjoyed the article except for 2 things. The link in Joel's RSS has an affiliate code, and they made me click on a link to go to page 2. Both of these suggest the kind of incentives he's railing against.
I'm suggesting that affiliate programs are the kind of incentive he's complaining about. And the artificial need to click to page 2 is even moreso - i'ts a direct attempt to inflate page views while reducing the quality of the customer experience.
Amazon Affiliates: get people to buy something they wouldn't have otherwise considered buying. I'm reading a book review and thinking about buying it, perhaps. But if I can be referred and the entire process takes two clicks, I'm much more likely to do it. Why pay? To give reviewers motivation to it; plus, that's the marketplace. You think Barnes & Noble don't have an affiliate program?
Affiliate code for inc: That's a tougher one. I doubt Joel actually gets paid for the viewers he brings to Inc, more likely there for traffic purposes. Paid affiliates to news sites kind of scare me, yeah.
2nd page: yeah, as a reader I LOATH this, especially now that NYTimes.com has removed the 'single page' button. Not really an incentive issue, though, is it? More of a trade-off between making money and pissing off the audience.
The problem in both cited cases was that the employee had the ability to give discounts. This is shocking. At most major retailers, only supervisors have that ability, and even there, they show up in audit forms that someone would see.
Neither is really a case of an incentive being gamed so much as lax security procedures. Simple auditing can sniff those out, and they're clearly violations of company rules over which the employee can be disciplined and probably fired, as per Joel's incentive plan.
AKA why you shouldn't evaluate testers based on the number of bugs fixed and developers based on the lines of code written (though I know places that still do both those things).
The story at the start of the article doesn't hold water with me.
If you went into a store and picked up a paid of $49.99 trainers (let's assume no sales tax), and you had this trick pulled at the register, but were then signing a receipt with $50.98 written on it, you're going to notice! Unless you're the sort of person who never looks at price tags or understands what the total should be, how would you fall for this?
But anyway, back to Austin, the Harvard professor. His point is that incentive plans based on measuring performance always backfire. Not sometimes. Always.
Always?!?
Normally, I like Joel's writings, but I have to say this is the dumbest thing I've ever read on hn.
I guess that's why Austin's a Harvard professor earning a living on the bequests of people who know how to implement incentive systems.
The pinnacle of this has to be the competitive incentive programs used at certain large software firms. In a competitive review system, the best policy is to passively sabotage your coworkers. This is robust even in the case where you are the rare superstar who has all the knowledge (in which case you simply hoard your knowledge). Those that try to be stars simply attract more coworker subversion. The countervailing factor is the general misery caused by this system. Employees will continue to sabotage until the work environment is as bad as they are willing to tolerate at the prevailing pay level. Thus, the only guaranteed result of a competitive review system seems to be maximally low morale and the majority of employee effort spent on gaming the system.
The late W. Edwards Deming (one of my heroes) proposed that the solution to this (in a manufacturing context) was to have a collective reward system (e.g., profit sharing), and to treat high and low performers as special cases. High performers should be examined to see if their techniques can be generalized across the organization, and low performers are examined to see if their performance is a result of the system or (rarely) a cause idiosyncratic to the low performer.
Many of the software types I've spoken to balk at such a method being applied to their organizations, because "creating software is not a production line process." This seems specious to me.
Have any of you if you've worked in a software house that had a reward system in place like the one Deming advocates?