How many Microsoft employees does it take to change a lightbulb?


UPDATE: This article was featured in The Best Software Writing I. Thanks Joel!

Joe Bork has written a great article explaining some of the decisions that go into whether a bug is fixed or not. This means that I can cross that one off my list of potential future entries. Thanks Joe!

But while I’m at it, I’d like to expand a little on what Joe said.His comments generalize to more than just bug fixes. A bug fix is one kind of change to the behaviour of the product, and all changes have similar costs and go through a similar process.

Back when I was actually adding features to the script engines on a regular basis, people would send me mail asking me to implement some new feature.Usually the feature was a “one-off” — a feature that solved their particular problem. Like, “I need to call ChangeLightBulbWindowHandleEx, but there is no ActiveX control that does so and you can’t call Win32 APIs directly from script, can you add a ChangeLightBulbWindowHandleEx method to the VBScript built-in functions? It would only be like five lines of code!”

I’d always tell these people the same thing — if it is only five lines of code then go write your own ActiveX object! Because yes, you are absolutely right — it would take me approximately five minutes to add that feature to the VBScript runtime library. But how many Microsoft employees does it actually take to change a lightbulb?

  • One dev to spend five minutes implementing ChangeLightBulbWindowHandleEx.
  • One program manager to write the specification.
  • One localization expert to review the specification for localizability issues.
  • One usability expert to review the specification for accessibility and usability issues.
  • At least one dev, tester and PM to brainstorm security vulnerabilities.
  • One PM to add the security model to the specification.
  • One tester to write the test plan.
  • One test lead to update the test schedule.
  • One tester to write the test cases and add them to the nightly automation.
  • Three or four testers to participate in an ad hoc bug bash.
  • One technical writer to write the documentation.
  • One technical reviewer to proofread the documentation.
  • One copy editor to proofread the documentation.
  • One documentation manager to integrate the new documentation into the existing body of text, update tables of contents, indexes, etc.
  • Twenty-five translators to translate the documentation and error messages into all the languages supported by Windows.The managers for the translators live in Ireland (European languages) and Japan (Asian languages), which are both severely time-shifted from Redmond, so dealing with them can be a fairly complex logistical problem.
  • A team of senior managers to coordinate all these people, write the cheques, and justify the costs to their Vice President.

None of these take very long individually, but they add up, and this is for a simple feature.You’ll note that I haven’t added all the things that Joe talks about, like what if there is a bug in those five lines of code? That initial five minutes of dev time translates into many person-weeks of work and enormous costs, all to save one person a few minutes of whipping up a one-off VB6 control that does what they want.Sorry, but that makes no business sense whatsoever. At Microsoft we try very, very hard to not release half-baked software. Getting software right — by, among other things, ensuring that a legally blind Catalan-speaking Spaniard can easily use the feature without worrying about introducing a new security vulnerability — is rather expensive! But we have to get it right because when we ship a new version of the script engines, hundreds of millions of people will exercise that code, and tens of millions will program against it.

Any new feature which does not serve a large percentage of those users is essentially stealing valuable resources that could be spent implementing features, fixing bugs or looking for security vulnerabilities that DO impact the lives of millions of people.

UPDATE: KC Lemson and Raymond Chen and Chris Pratley have opinions on this as well.

Comments (51)

  1. Robert Sayre says:

    What a fantastic argument for Open Source!

  2. Raymond Chen says:

    Who develops the test plans for open source software? Who updates the screenshots in the user’s guide and online help? And who translates the documentation into Polish and Turkish? Who verifies that the feature doesn’t violate the Americans with Disabilities Act or German privacy laws? (Back when I worked on Linux, the answer was "Nobody. There is no test plan, there is no printed user’s guide, what little documentation there is exists only in English, and nobody cares about complying with the ADA or German privacy laws." Maybe things have changed since then.)

  3. Eric Lippert says:

    I’m not following you Robert.

    Thought experiment: tomorrow, Bill open-sources all Microsoft products and set up a hundred billion dollar endowment fund to pay for continued development of the codebase. What makes all of the costs I mentioned suddenly go away?

    Nothing! Open source isn’t magic. There seems to be a strange belief amongst the open source community that just because you can make a change to the source code, and no one pays you to do so, that the change was free. But it wasn’t, because changes don’t cost money. Changes cost EFFORT to do right, and money is just a convenient way to measure effort, not effort itself. There’s only a finite amount of effort in the world, and knowing how to apply it to greatest effect is a difficult problem.

    It doesn’t take a couple dozen people to change a lightbulb here _because_ we sell software for a living — it would take those couple dozen people even if we gave it away with the sources. It takes a couple dozen people because we deeply care about that legally blind Catalan-speaking customer. It takes a couple dozen people to change a lightbulb because software is an insanely complicated device that runs in an insanely complicated world. Managing that complexity is a lot more work than changing the code.

  4. Robert Sayre says:

    The point is that the customer is free to fix/alter that which affects themselves, but isn’t worth fixing to you. Since the feature/problem isn’t worth it to you, it obviously follows that the feature isn’t for widely deployed software… where all of your effort is (rightly) focused. It just needs to work on a small number of boxes for a very specific purpose, where the blind Catalan-speaking customer doesn’t come into play. Or perhaps the feature is very very important to the customer (yesterday! dangit!), and they don’t have time to wait for MS to QA on Windows XP Home Arabic.

  5. Eric Lippert says:

    Got it. But this is precisely why we built the script engines to be extensible by arbitrary third party ActiveX objects. Like I said, if the feature is cheap and easy, then implement it yourself in a VB6 object. But the script engines themselves cannot be allowed to fracture into a million slightly different mutually incompatible versions — that doesn’t serve customers well.

  6. Ian Ringrose says:

    Now what if you ‘shipped’ an “open source” set of helper ActiveX objects for the script engine. You could post it on one of the Microsoft web sites and just say that it is a demo of how to extend the script engine… Can we find a modal that lets Microsoft employees get out ‘quick hacks’ without Microsoft being responsible for then in the nest 100 years?

    No all software help to be written to the same standard…

  7. Anonymous says:

    I believe MS already ships quit a significant amount of "open source" software in the form of all the samples in the msdn online, and the samples included with the various product installations, such as VB.

    And if that is not enough, you can find tons of samples online, in various trade publications, in newsgroups and obviously in blogs like this one.

    OTOH you cannot ignore the fact that Microsoft becomes somewhat liable for any piece of software it ships regardless of how unofficial it is. If you copy a sample for this site and it formats your disk, you will blame Microsoft and Eric could loose his job.

  8. Dan Shappir says:

    BTW, I wrote the above comment.

    For some reason, the "Remember Me" check box doesn’t work for me. And I got a memory exception screen when I submited.

    More untested, buggy software from Microsoft 😉

  9. Michael Howard says:

    You missed one step – three people to argue about whether this method is safe for scripting :-)

  10. Anonymous says:

    Raymond wrote: "Who develops the test plans for open source software? There is no test plan, there is no printed user’s guide, what little documentation there is exists only in English, and nobody cares about complying with the ADA or German privacy laws." Maybe things have changed since then"

    They certainly have changed. If you’re interested in whether this is true, check out the Gnome and KDE projects. Much more organised (esp. KDE) than they ever used to be. In general it’s work that most OSS developers don’t want to do – it’s not an itch to scratch to use the cheesy phrase I keep hearing. Now that there is interest in OSS and money is being thrown about, people are working on these things because they are being paid to do it.

    Eric seem to be arguing (please correct me if I am wrong) "How do we pay for peoples effort if we give everything away"? Well for a start OSS does not mean free. Just because some OSS apps are free doesn’t mean they *have* to be.

    I wonder how many Microsoft employees have actually ever been to opensource.org and can now argue about OSS *without* resorting to the usual FUD we get from Microsoft. I’m not deliberately trolling here, I like OSS and I also like Windows, but when I see the FUD and things like the underhand funding of SCO I don’t like MS. It’s like they don’t want to compete on fair terms – maybe I’m just naieve and that’s how business works. I hope not.

    I have to say about the user guide and it being printed, I have NEVER seen anyone read it. No exageration. Never. Ever. Maybe you should do a poll, you could save money by not printing it :)

  11. a. says:

    geeks doesn’t read it. joe user does.

  12. Eric Lippert says:

    > Eric seem to be arguing (please correct me if I am wrong)"How do we pay for peoples effort if we give everything away"?

    No, I’m certainly not arguing that. Perhaps Raymond is, but I am not. I realize that there are companies that pay people to work on open source.

    My argument is that IF you are in the business of writing software that is to be used by millions of people around the world, THEN the primary cost of implementing that software is NOT in the implementation. The cost is in the design, the review, the documentation, the testing, the maintenance, the support calls, etc.

    Whether your business model calls for selling that software or — as we did with the script engines for the last seven years — giving it away for free, whether the source is open or closed, has not the least bearing on my point. Does the open source model work well for one-off changes that will be distributed to one person? Obviously. But that’s not the kind of change I’m talking about.

  13. Dan Isaacs says:

    Opportunity costs are hypothetical. "Stealing" refers to the act of depriving someone of actual property. As a hypothetical is not actual property, "stealing" is a poor word to use in your last paragraph. Better to let the facts stand up for themeselves instead of concluding with a false characterization.

  14. Eric Lippert says:

    > Better to let the facts stand up for themeselves

    Actually, "facts" are abstract entities which do not have legs. So they can’t really "stand up" for themselves. That’s kind of a poor choice of words, wouldn’t you say? Fortunately, by using my advanced skills in inference I can probably figure out what you intended by your imprecise and colloquial expression.

    :-)

  15. Andrew says:

    You can tell that these are only techs reading this because hey guys! (particularly Robert, etc) – End-users don’t want half-baked software!!!

    While as a dev, I would gladly rather have the software do what I wanted it to, I would also rather have it do what it can do (without causing more problems), and allow me to change its behaviour.

    Just because Tech Guy down the street can change the behaviour of an application doesn’t mean that all end-users can. We, in the tech community, are in a position where we not only "see" the final result but also see how its made and can affect change. End-users, although this attitude is sometimes changing, don’t care how it works – they just want it to.

    When I ask the time, I don’t want to know how to build a clock nor how to fix the clock – I just want the clock to tell me the time. Open Source isn’t a solution for the problem mentioned above- it’s simply another take on it.

  16. Anonymous says:

    "Just because Tech Guy down the street can change the behaviour of an application doesn’t mean that all end-users can."

    Admittedly not, but at least they would have the opportunity to pay/hire someone to do it for them. If I want someone to add something to my house heating system, I want *ANYONE* with the skills to fix it to be able to fix it, I don’t want to be FORCED to go back to the original installer ….

  17. Rj says:

    So the Answer is "41 at a minimum"?

  18. It looks like the intricacies of higher-dimensional geometry will have to wait another week; I am incredibly…

  19. John Gruber makes an appearance in the soon to be released book The Best Software Writing I which was put together by Joel Spolsky .

  20. Whether it’s software or content, making a change can often be a much bigger deal than you’d think. I…

  21. Patrick Schmid says:

    Eric,

    just wanted to make sure you saw my response to your comment on my blog:

    well deserved feedback. I remember vividly that I made the argument that a certain bug fix for Outlook 2007 would only be a few lines of code (I stopped short of volunteering to write the code for the devs) without knowing how much other work would be associated with those few lines.

  22. David says:

    What if the user can’t create an ActiveX object?  Not becauser they can’t code, but because they don’t have access to the development environment, compiler, or sufficient authority to create and/or install the object?  After all, VBScript (and JScript, VBA, WSH etc etc etc) is available to many more users than visual studio, or some other development tools.

    You can argue that they should obtain the neccesary software/authority etc to develop ActiveX objects (and perhaps after 12 months the business case might get approved…).  But a language that requires a business case to implement some needed functionality is essentially crippled.

    I can say from personal experience that having a language not go "all the way" is as frustrating as the allusion implies.

  23. RK says:

    So much about open source talk. Forgive me for my ignorance. Who uses open-source? I have heard everyone promoting it. Haven’t seen anyone, at least in the develoment community that i have seen till date using any of those products. Let’s not be hypocritical. You want to make something that people use, it costs. Listen nothing comes FREE.

  24. Yuhong Bao says:

    Sounds so similar to bureaucracy!

  25. Blah says:

    As much as I love Open Source software, this arguement (the one made by the article-writer) holds true for Open Source software, too.  However, with Open Source software, you usually don’t have the luxury of a Program Manager, multiple QA folks, etc all collaborating.  It’s usually you and maybe a couple other dedicated folks passionately working on something.  You get done, you toss it out to users, and then tons of bugs show up that you never thougt about (who’d have thought the user’s would stick the bulb outside in zero-degree weather…OOPS!)  So, in my opinion, this is not an argument "for" Open Source software.  Open Source software suffers from the same "how many does it take" syndrome.

    Getting back to the article writer’s point, you don’t want to toss all kinds of one-off crap into your project, because then it bloats up the project.  However, more and more projects opt for a foundation and then extensible scripting (EG: some video games, firefox, etc).  Of course, if your project is nothing more than a scripting language (VBScript), then it’s meant to be light-weight.  So, yeah, suck it up and do the 5 lines of code yourself.  If VBScript got bloated with everyone’s "one-off" junk, it’d be such a cumbersome hodge-podge of stuff, nobody would use it (or complain about how complicated it’s gotten.)

    I think Bill Cosby said it best…"I don’t know the secret to success, but the secret to failure is trying to please everyone."

  26. Blah says:

    Lots of folks use open source.  From the person who decides to use Open Office or Firefox, to the person who downloads a piddly little program or trainer to hack some game file so they can buff up their character.  Software is so ubiquitious and easy to make these days, that folks think it has to cost tons of time and money to make it still.  Open Source just means the source is viewable and (possibly) modifiable (depending on the license the author releases it under).   Open Source isn’t necessarily free in some cases…folks can let you see the source, but still charge you to use it or the compiled program.  But, software isn’t some commodity large corporations have sole entitlement to.  It’s like regular writing…anyone can do it, and lots of folks do.  Whether you choose to buy a book from the store that has the info you want, or get the info for free from an internet site is a matter of choise.  Free doesn’t necessarily mean "bad" or "poor quality".  However, in United States especially, folks think the more something costs the better it is.  So, there’s still a frown on Open Source and Freeware.  It’s like saying only the expensive Dealer can fix your car, because the shade-tree mechanic down the street doesn’t know anything.  On the contrary, the shade-tree mechanic may know a lot.  Then again, he may not.  There’s a greater variance in quality with the shade-tree mechanic, but you can still get poor service from the Dealer.  Quality is not a given just because you pay money for a product or service.  And, inversely, poor quality is not a given just because you get something for free.

  27. It looks like the intricacies of higher-dimensional geometry will have to wait another week; I am incredibly

  28. bob says:

    I think the problem is that the development tools are not free.   Writing and ActiveX extension as the author suggests shows great extensibility, however, if I have to spend $500 or $1000 to get the tools to do this then it’s a non-starter.

    Just one more reason Mac OS X with free developer tools is superior. :-)

  29. The problem that Microsoft has is often scale!  When you have millions of people that depend on your project changes that normally are done without a second thought are now all of a sudden very difficult.  This is the huge dependency on product use.  Divide the number of people using your product by the number of people making your product and you might get a better idea of the efficiency of your team (not the best metric I know but it helps with perspective).  The second problem that is described is flow.  Most of the time described in the many week process is waste due to time waiting on other people.  If you had everyone in the same room then the task would be completed much faster.  Open Source is great, but Open Source has the same problems at larger scales.  These scale issues often cause the community to "fork" and that can make this better but can also cause fragmentation.

  30. anonimous says:

    It doesn’t take a single employee to change a lightbulb, Bill Gates just redefines ‘Darkness’ as the new industry standard…

    (it was in my englishbook bill…)

  31. Aakash says:

    Love it!

    I tired telling my managers how much time a little bug may take to fix and get on production. :-)

  32. Matthias says:

    I can see the point you’re making in the article. In fact this is why I value stuff coming from Redmond so much. It actually is tested, some API concerns are taken into account and above all one can actually find some real-life examples of how to use a particular part of the system.

    No wonder it takes so much to implement such a trivial thing.

    A note about open-source frameworks: they are great, even fantastic! But the best ones have extensive documentation and a fat set of examples. The ones that don’t have that learning resources, even if they are best in the issues they are solving, are doomed to be used once, maybe twice.

  33. Zbyszek says:

    I totally agree with this article. People too often do not realise that a simply change can cause such a huge side effects. Our website is in 16 languages, we have accepted that some text (on some pages even majority) is not translated, because traslating it caused more grief than happines. For example translating some text caused grammatical errors (consider plural form – in English is easy, 1 photo, 2 or more photos, buut in Polish is 1 zdjecie, 2 zdjecia, 5 zdjec, 12 zdjec, but .. 22 zdjecia, however 21 zdjec, etc) and caused a flood of complains, which created an extra task of answering them, which … costs time==money

    @bob – of course you have free development tools (on MS platforms – Express editions, on Linux Eclipse, etc)

    @open source supporters – I love some opensource stuff, but you must accept that if someone makes changes which are just for him they have to accept that the WHOLE modified product cannot be then properly supported, as their small change in the code might have a knock off effect in a very unexpected place. I have developed many libraries in my life and I know from the painful experience that you will get errors in totally unexpected places.

    One the strangest erross we have had had a following scenario (code written in C, many years ago). User input was accepted as a number, including decimal comma. Then we introduced localization allowing decimal comma, routine was scanning a number changing comma to dot, like while(*ptr!=’,’) ptr++; *ptr=’.’.

    Of course once it happenned that users input did not have a comma and code (it was C!) was continuing happyly beyond array and … found value 2c on the stack and converted it to 2e. It happenned to be return address, so as the effect next line after the call of this routine was not executed, and this skipped was a short jump out of the loop, so external loop executed once more and customer name was changed to something else. It took us almost two days to find this problem.

  34. Rachel 'Groby' Blum says:

    It's a classical example of how when a company gets bigger, it starts moving slower and slower (by necessity. MS ignoring any of the above points will almost guaranteed result in a lawsuit somewhere)

    Reading this, it strikes me that the software industry pretty much as a whole has no way of releasing features "along the way" to get better feedback. There's no way to release a simple, non-localized, non-vetted LightbulbEx, collect feedback, improve it, and only finalize it later – if it's indeed a crucial feature that everybody needs.

    And while I love OSS, it's not exactly the answer. You still need to move from that quick-hack version to the fully localized one, you have the issue of incompatibilities between different versions of your SW (which non-geek users don't exactly appreciate), etc.

    It's an interesting problem that hasn't seen a solution yet.

  35. bolthar says:

    Maybe it's time to change the way you guys release your products – not that I know *how* you should change…

    I'm not following your line of thought. How does changing the distribution model affect the costs of design, implementation, testing, security review, documentation, translation, maintenance or management? Distribution cost is a cost that I did not even think to mention, so I don't understand why you're bringing it up. Can you explain? – Eric