• Redeemer (unregistered) in reply to Mel
    Mel:
    fred:
    The worst of all occurs when management itself start gaming the system (for instance, if their own bonuses depend on how small the bonus for their people are -- don't laugh, I have seen that).
    WTF????? So the team does a crap job, and the manager is rewarded for that? [sarcasm] Now that'd be a great place to work...[/sarcasm]

    I think what fred tried to say, is that since the managers get a bonus based on the employee bonus, they would give them a higher bonus.

    Though on re-reading the post i am not that certain of the meaning. But i thought i'd still put in my 2 cents

  • Ned Batchelder (unregistered)

    This very idea was the subject of a Dilbert cartoon years ago. When Wally hears the news, he proclaims, "I'm going to code me up a minivan!"

  • Flubu (unregistered)
  • Flubu (unregistered) in reply to Flubu

    How sad is it that I'm hotlinking from my own damn website :)

  • ClaudeSuck.de (unregistered) in reply to A Nonny Mouse
    A Nonny Mouse:
    so.... they got paid extra to do their jobs... i see..

    No they got paid for taking out bugs they had produced before while getting paid already. This is like a dog who's chasing his tail.

  • (cs) in reply to Jason
    Jason:
    At work we have a 'beer fund' that you are expected to contribute to if you break the build -- IF you check something in that causes other people to have to fix your bugs to work.

    Nice. Note the difference with the WTF: voluntary, apparently set up by the developers themselves, and used to subsidize a social event that (I guess) all the developers would attend.

  • Winslow Theramin (unregistered)

    There should be a poll on most of these stories where readers get to vote on whether or not they believe the story is real or just made up. This doesn't sound believable at all.

  • Cpt (unregistered) in reply to Winslow Theramin
    Winslow Theramin:
    There should be a poll on most of these stories where readers get to vote on whether or not they believe the story is real or just made up. This doesn't sound believable at all.

    Wake up, welcome to reality, sadly there are stupid CTO's, and frauduleous developers and testers out there.

  • JM (unregistered) in reply to Winslow Theramin
    Winslow Theramin:
    There should be a poll on most of these stories where readers get to vote on whether or not they believe the story is real or just made up. This doesn't sound believable at all.
    The fact that there was a Dilbert cartoon about it (posted *twice* now, since nobody's interested in other people's comments) pretty much proves it really happened. As Scott Adams himself remarked, he's never drawn a cartoon that didn't result in mail not only from people who had this exact scenario happen, but people who can top it with something even more ludicrous.

    Aside from that, the only thing that's required for this is stupid management, which has grasped the concept of rewarding people but is not quite able to foresee the consequences of their harebrained schemes. How exactly is that unbelievable?

  • ClaudeSuck.de (unregistered) in reply to Pete Jackson
    Pete Jackson:
    It sounds like this program was successful only because it didn't last very long. There is fallacy in expecting "A" and rewarding "B". In this case, the CTO expected quality, and chose to get it by rewarding developers and testers for "fixing bugs".

    Much more difficult is finding a way to reward developers and testers for submitting bug-free code to the master sources. That's not easy.

    Of course, one could argue that any time you choose to use extrinsic motivators to get results that should be expected as part of someone's job, you'll have people who game the system.

    My advice? Find some quality measurements and have them delivered to the team periodically. Don't let the bosses see them. If the team is intrinsically motivated, they'll improve quality without you having to pay a bug bounty.

    bug-free code??? Never seen this! Where can I get some? plz send me ze bug-free codez

  • ClaudeSuck.de (unregistered) in reply to Alin
    Alin:
    Code Slave:
    OB Dilbert: I'm going to code me a mini-van after lunch!

    I was thinking more along the lines of a Corvette... hmm... need more bugs!

    bugs = bucks, that's it!

  • Hova (unregistered)

    Outlaw, how could it be successful? Did you miss the part about developers making bugs on purpose so they could get paid to fix them?

    Although I like the premise you'd have to find a way around abuse of the system...

  • (cs) in reply to Mike Arthur
    Mike Arthur:
    Am I the only one that sees deliberately introducing bugs just to fix them as being massively unethical?

    I don't care if your coworkers are doing this, it's well out of order in my opinion...

    Like sleeping on the job when you're being paid by the hour?

  • (cs) in reply to Seth
    Seth:
    The actual solution is a slight variation on a previous suggestion. There will be bonuses totalling $1000; testers get $10 for every bug they find, $10 is taken from the pool for every bug not fixed in a certain timeframe, programmers get the remainder.

    The actual solution is that you let the developers and testers do their jobs for the salary you pay them. If they don't do a good enough job, replace them until you find people that will.

  • (cs) in reply to bohica61
    bohica61:
    I just interviewed at a company where they asked me, "Do you have any problem with working 48 hours a week? Because that's what's expected."

    Sure, as long as you don't mind my salary requirements going up by 20%.

    I have interviewed at a couple of companies that basically explained to me that they were in a perpetual state of deathmarch. And they wonder why I didn't accept the positions they offered.

  • facilisis (unregistered) in reply to Leo
    Leo:
    I can't imagine working in such a poisonous atmosphere that every gesture of goodwill by the company, every attempt to make things better is immediately turned around into a way to screw that company.

    I guess that make you unemployed then?

  • Ilya Ehrenburg (unregistered) in reply to KenW
    KenW:
    Seth:
    The actual solution is a slight variation on a previous suggestion. There will be bonuses totalling $1000; testers get $10 for every bug they find, $10 is taken from the pool for every bug not fixed in a certain timeframe, programmers get the remainder.

    The actual solution is that you let the developers and testers do their jobs for the salary you pay them. If they don't do a good enough job, replace them until you find people that will.

    Should only work if you pay an adequate salary, which is not too common.

  • (cs) in reply to Redeemer
    Redeemer:
    I think what fred tried to say, is that since the managers get a bonus based on the employee bonus, they would give them a higher bonus.

    Though on re-reading the post i am not that certain of the meaning. But i thought i'd still put in my 2 cents

    No, what Fred said is that he's seen cases where there was a pool of bonus money given to a manager, and the manager made sure that there were enough things blamed on the programmers to justify reducing their bonuses, thereby increasing the amount remaining in the bonus pool that the manager could allocate to himself.

    I've seen this as well, although never to an extreme.

  • Mark (unregistered)

    They should adopt a make or take model.

    The tester gets $15 per bug found. The programmer pays $10 towards that.
    He then gets $25 back once it is fixed.

    Bugs and testers should be randomly allocated...

  • (cs) in reply to Mark
    Mark:
    You could have a secondary market allowing other developers to guess the future price of a fix, based on their assessment of the complexity. For instance, developers decide that the Paula Bean module rewrite will take 6 months, but another developer believes the project will be cancelled, and so is going to be 'short' on that bug.

    In all seriousness, if they're worried about developers/testers gaming the system, they could have a weighted system such that the bonus diminishes with the number of bugs fixed. That way, the cost of creating bugs becomes less than the bonus for that bug. The age of the bug would be a weight, since the age of a bug corresponds roughly to it's difficulty and importance.

    Analysis like this may well be the fundamental underpinning of the CDO fiasco.

    It took me all of thirty seconds to ask myself how to game such a system. If I'm part of a team implementing, say, a six month long project, all I have to do is to spend two weeks generating bugs, five months on the functionality, and then offer the fixes in the last two weeks. A certain amount of violence with prejudice may be necessary to stop the testers from fixing anything before then, but in all honesty even this can be solved trivially -- they can fix the bug whenever they like, as long as they don't report the fix.

    In a principled world, I'd prefer that management concentrates on encouraging useful software (commercial or otherwise) rather than silly fucking games. A decent ANTLR parser for C++ would be nice, if only for the tool-building opportunities provided by the accompanying AST. An OSS customer relationship management system would be excellent. I'd like a proper state-driven framework for the V bit of MVCs, as well, but then I'm barking at the moon. Well, we can all dream; meanwhile, we get this sort of rampant silliness.

    In re the Winston Churchill paraphrase, I think that George Bernard Shaw is perhaps more pertinent: when a woman wrote to him asking to bear his child -- "think of the child with your brains and my beauty!" -- he replied "But what if he were to have your brains and my beauty?"

    As far as I am aware, the Law of Unintended Consequences is the only law that has no known exception to prove the rule.

  • (cs) in reply to Winslow Theramin
    Winslow Theramin:
    There should be a poll on most of these stories where readers get to vote on whether or not they believe the story is real or just made up. This doesn't sound believable at all.
    You've been waving your surname too close to a van der Graaf generator, haven't you?
  • The National Bug Association (unregistered) in reply to dlikhten
    dlikhten:
    Testers should not be measured by the number of bugs they create, and developers should not be measured by the number of bugs they fix.
    Testers don't create bugs. Developers do.
  • John Smith (unregistered) in reply to Outlaw Programmer

    Except, if this were the case, all the developers would make sure they didn't close off the last issue. Therefore, they would still qualify for the bonuses as they generated more (artificial) defects to refill the queue.

  • gwern (unregistered) in reply to Danny V

    Unlikely; by his universality test, this would fail - if everyone cheated, then the 'market' would soon collapse or be canceled when it became obvious, despite the long-term reward of always having a market (equivalent to a pay raise). It's the Prisoner's Dilemma again - short-term, defecting/cheating is the thing to do, but unfortunately, we want to live in a long-term world....

  • tech support (unregistered)

    Since nobody has mentioned it yet, I'll point out that there is a Dilbert cartoon about this exact topic. Here is a hotlinked copy of the cartoon.

  • w (unregistered) in reply to DOA

    Agreed on the compensation in general. Any system you come up with, developers and other smart people will game and it will be counter productive. Nothing at all will work in the long run. The key is, 'in the long run'.

    But this CTO? It sounds like he had planned right from the start for this to be short term. As a one time thing, and this could only work once, it looks like it worked rather well.

    It got the QA and developers conspiring together, they squashed lots of bugs in handful of days. The cost to the company is negligible. It was killed long before any real damage to the product followed.

    I'm impressed. Of course, now that we've all read about this, no one else can pull it off.

  • Leg Humper (unregistered)

    I'd love to work for a company where the employees all pair up. I just need to find out if BustedTees.com is like that so I can pair up with Irish girl :D

  • Topher (unregistered)

    Dilbert covered this idea years ago. The last panel had the developer saying "oh boy, I'm going to go program myself a new minivan!"

  • Jakob Engblom (unregistered) in reply to ObiWayneKenobi
    ObiWayneKenobi:
    Glenn:
    I swear I've heard this story before.

    Probably because it was printed in Redmond Developer magazine or whatever it is that Alex writes for. I recall this one as well.

    And I heard it from a different source several years ago. I fear that this thing has been tried by many companies at different times, as it looks obvious as a "good idea".

    If you want to know about broken incentive systems, read some deep book on Soviet economic history. They tried all kinds of things similar to this to make people do the right thing within the frame of a command economy. Suffice to say that it did not usually work too well.

    The only thing that really works is a motivated workforce who feels that they should do the right thing based on overall targets. But that is harder to manage into existence, requires investing in your people and building a sound culture.

  • Fire! (unregistered)

    The real WTF is that Damon still has his job. After Damon showed a sense of ethics and refused to be a team player, there were TWO glaring reasons to show how he couldn't get along with the rest of the company.

  • Hans (unregistered) in reply to Mike Arthur
    Mike Arthur:
    Am I the only one that sees deliberately introducing bugs just to fix them as being massively unethical?

    I don't care if your coworkers are doing this, it's well out of order in my opinion...

    You are right, but I'd feel worse about it if companies weren't such complete psychopaths themselves.

  • Ant Mitchell (unregistered)

    History Shmistory - right ?

    http://www.robweir.com/blog/2007/05/legend-of-rat-farmer.html

    Question is did the CTO know this, or not. I suspect he did.

  • Richard C Haven (unregistered) in reply to Danny V
    Danny V:
    The REAL WTF is that Damon felt that there was an ethical problem with taking advantage of an absurd system of compensation!

    No, it is an ethical problem. You can choose to cheat your employer by failing to do your job as best you can; or you can choose to watch others benefit by cheating without doing it yourself. It sort of depends on how you view your own life.

    Cheers

  • Bowie (unregistered)

    They should have offered $1000 to anyone who could have pointed out how damn dirt simple it would be for people to cheat the system back when they originally proposed it.

    I woulda made a quick G on that one. :)

    In my experience, the best developer<--->tester environments are deeply, deeply adversarial. Make them hate eachother, and your code will be absolutely indestructible.

  • Bernd Wechner (unregistered) in reply to Outlaw Programmer

    Stuff and nonsense. If that became a cycle, then the team could design in bugs from the outset knowing to expect the defect market. It can only work by total surprise with no expectation and hence only work once, and doing it the once has the possible consequence that on the next release staff, suspecting the manager thinks as simply as you just have, expect a new open defect market as the release stabilisation phase approaches and build in the bugs they need for extra dosh and the defect market never opens so they ultimately just added significant delivery delays ...

    I suggest a read of Robert D. Austin's Measuring and Managing performance in Organizations to get a handle on how WTF this really is and flawed logic. It embraces one short term success aginst medium term and onward dysfunction (Austin's term).

  • Iwein (unregistered) in reply to Outlaw Programmer

    If the benefit is not directly related to the health of the end product, but to the perceived process needed to get it in place, self learning agents will find the easiest way to game the system.

    The more I think about it the more retarded this path looks to me. If you pay bonusses for bugs you will get more bugs. If you only pay bonusses for more than 50 bugs you will get more than 50 bugs.

    How about this scheme: Every release testers and developers get a bonus, the bonus is devided by the number of production bugs found in the first x time of operation.

    My captcha reads: "validus"... how about that?

  • howlingmadhowie (unregistered) in reply to keith

    horrible idea. most people work worse under threat of punishment, not better.

Leave a comment on “The Defect Black Market”

Log In or post as a guest

Replying to comment #:

Log Out

« Return to Article