Reply to: Somebody's Law

[RSoles]

Goodhart eh?
Thanks all.

[vetran]

nothing wrong with Prince 2 etc, properly implemented.

I used to get a bonus based on Metrics & individual targets as did the rest of business. at the end of the quarter the number of IT calls would go up dramatically because people realised their bonuses were at threat and they had to do something,anything. That normally meant leaning on IT then trying to blame them. Had a few irate managers complaining at me because we had made their staff miss their bonus, I had to show them the call and explain it was 3 hours old. Last week of the quarter was affectionately known as 'get me my bonus week' in IT. However things got done and money was saved.

Now however we get a percentage depending on overall pay and the profitability of the whole EMEA. Something we have no control over.
Now people don't care less. No spike at quarter end.

The difficulty is that either people don't spend the relevant time to do the process properly or are so anal about it they cause people to subvert the system. Any gaming of the system would be 'gross misconduct' so the manager needs to grow a pair and watch carefully. They will only need to publicly flog one developer. Probably the worst one as they won't have the brains to cover their tracks.

[IR35 Avoider]

There's a law that states whenever a performance metric becomes a payment target it loses its value as a metric.

Anyone recognise this? point to a source?
Ta

Goodhart's law - Wikipedia, the free encyclopedia

Goodhart's law, although it can be expressed in many ways, states that once a social or economic indicator or other surrogate measure is made a target for the purpose of conducting social or economic policy, then it will lose the information content that would qualify it to play that role. The law was named for its developer, Charles Goodhart, a former advisor to the Bank of England and Emeritus Professor at the London School of Economics.

The law was first stated in a 1975 paper by Goodhart and gained popularity in the context of the attempt by the United Kingdom government of Margaret Thatcher to conduct monetary policy on the basis of targets for broad and narrow money, but the idea is considerably older. Closely related ideas are known under different names, e.g. Campbell's Law (1976), and the Lucas critique (1976). The law is implicit in the economic idea of rational expectations. While it originated in the context of market responses the Law has profound implications for the selection of high-level targets in organisations.[1]

It has been asserted that the stability of the economic recovery that took place in the United Kingdom under John Major's government from late 1992 onwards was a result of Reverse Goodhart's Law: that, if a government's economic credibility is sufficiently damaged, then its targets are seen as irrelevant and the economic indicators regain their reliability as a guide to policy

[Mich the Tester]

Joel Spolsky ranted about this IIRC but he didn't mention any kind of law, just that developers will learn how to abuse the system... writing more verbose code to get more lines/day, checking in every save to boost commits, deliberately introducing bugs to increase number fixed, etc.

Yep; years ago I was on a test team where they gave bonusses for number of issues raised. Effect on quality was disastrous as issues were held back in early phases so they'd lead to bigger problems in later phases of testing.

[d000hg]

There's a law that states whenever a performance metric becomes a payment target it loses its value as a metric.

Anyone recognise this? point to a source?
Ta

Joel Spolsky ranted about this IIRC but he didn't mention any kind of law, just that developers will learn how to abuse the system... writing more verbose code to get more lines/day, checking in every save to boost commits, deliberately introducing bugs to increase number fixed, etc.

[original PM]

We currently have a situation where we cannot get marketshare data for certain areas of the country and so there is no local marketing activity going on.

But my argument is that we know what marketing activities have an impact in some areas of the country and as the business is pretty niche then you could with a high percentage of success implement these activities in all areas and you would see a return.

The counter argument is that if you cannot directly measure your marketshare there is no point in doing the local marketing activities.

So in those areas of the country where they can get marketshare figures they are targetted (lets so to obtain 38% marketshare) and as soon as they do they stop the marketing activity and low and behold next time they look they are back to 36% marketshare.

Another example of where KPI's have been used to drive a workstream when in effect that workstream should be part of BAU...

oh well best get back under my rock...

[EternalOptimist]

Actually bonus schemes and KPI's can work brilliantly well, if they are policed.

No one ever puts a cost on this or implements it (well..not many do)

[Mich the Tester]

Nothing wrong with KPIs its the way you implement them that is the issue.

Isn't that what insultants say about PRINCE 2, T*Map, ITIL, ASL, BISL, TQM and all the other big piles of expensive sounding tulip that occupy bookshelf space in corporate offices?

[vetran]

Nothing wrong with KPIs its the way you implement them that is the issue.

You could do it the right way:

Days on hand + Days to ship joint target= good KPI

or the wrong way

Days on Hand controlled by purchasing.

Days to ship controlled by warehouse.


or you could do it the Six Sigma brown belt way. (they are so full of it you need a belt to keep it in.)

Mark as 100% and de-rate when you get caught.

[bless 'em all]

If you're going to talk about corporate cackness, you cannot leave out KPI's.

I forgot about those ... just goes to show you CAN forget what it's like to be a permie.

[EternalOptimist]

I worked as a BA for a large service company, and one of the metrics was cost of parts fitted during servicing and repairs.

The idea was that engineers would love to fit loads of parts because as well as guaranteeing a fix, they are driven to do a prefessional job - they like perfection.
Of course the business cannot afford to spend a hundred quid on a machine that only earns 50 quid a year. So there is a tension there, and the metric is used to make sure the spend is neither too high, nor too low.

Then it was decided to include the metric in the bonus calculation, spending 90% of the budget giving maximum bonus. The idea being that this would provide an incentive for the engineers to toe the management line.

Costs to the business in spare parts went through the roof.

What happened was that the engineers went from having an aspiration to get the spend somewhere between the lower and upper limits, to avoid a bollocking from the boss, to having an imperitive to spend exactly 90% of the budget, to avoid getting a bollocking from the wife.

So in the quiet times, parts would be stored in garages or garden sheds, and booked off as being used. Some of these guys ended up with tens of thousands of pounds worth of gear, all going rusty or mouldy in the damp garage.

Plus every time the guys went off patch, they would book off loads of parts that were not actually fitted, so the target team picked up the costs, often ruining their bonus prospects, causing emnity, ruining and falsifying service histories. Eventually teams would not accept help from each other, so the management had to recruit more heads to cover the busy periods.

A simple little idea, designed to save a few hundred thousand quid, nearly brought that company to its knees

[Mich the Tester]

One springs to mind where employee satisfaction is a contributor towards the EOY bonus, along with some other tulip like profitability and sla's.

You have a situation where an 'employee' has to work harder for the company to hit the profit/sla/tulip %age then give a higher satisfaction score as an employee to hit the 'happiness' target or risk missing the bonus.

If you're going to talk about corporate cackness, you cannot leave out KPI's.

[bless 'em all]

There's a law that states whenever a performance metric becomes a payment target it loses its value as a metric.

One springs to mind where employee satisfaction is a contributor towards the EOY bonus, along with some other tulip like profitability and sla's.

You have a situation where an 'employee' has to work harder for the company to hit the profit/sla/tulip %age then give a higher satisfaction score as an employee to hit the 'happiness' target or risk missing the bonus.

[EternalOptimist]

There's a law that states whenever a performance metric becomes a payment target it loses its value as a metric.

Anyone recognise this? point to a source?
Ta

aha, bonus schemes

this is my specialist subject. I am going to love this thread

[Spacecadet]

There's a law that states whenever a performance metric becomes a payment target it loses its value as a metric.

Anyone recognise this? point to a source?
Ta

Sounds like the law of unintended consequences