It reminds me of the orange juice test.
You organize an annual convention for hundreds of people.<p>You tell the banquet manager of the hotel you are considering that the morning breakfast must include a large glass of freshly squeezed orange juice for everyone of the attendees. It must be squeezed no more than two hours before the breakfast.<p>It is not possible to do so. Squeezing that much orange in much a short amount of time would be prohibitively expensive.<p>If the manager says yes he is either lying or incompetent and you'd better find someone else who will tell you it's not possible.
This seems like a good opportunity to post RMS' rider again:<p><a href="https://secure.mysociety.org/admin/lists/pipermail/developers-public/2011-October/007647.html" rel="nofollow">https://secure.mysociety.org/admin/lists/pipermail/developer...</a><p>Fabulous stuff. I wonder how often he has/gets to hang out with random parrots since this document became widely known. In my mind he is surrounded constantly by sandal-wearing acolytes wielding exotic birds of every variety.
Engineer's version: Write-only memory<p>"
Out of frustration with the long and seemingly useless chain of approvals required of component specifications during which no actual checking seemed to occur, an engineer at Signetics once created a specification for a write-only memory and included it with a bunch of other specifications to be approved. This inclusion came to the attention of Signetics management only when regular customers started calling and asking for pricing information. Signetics published a corrected edition of the data book and requested the return of the 'erroneous' literature."
One of the interesting things about tech work is that it's almost all "brown M&Ms". It's amazing how important attention to detail is in this field and how quickly something will simply not work if the details aren't sweated.<p>We see it time and again when things go into production where the "brown M&Ms" haven't been looked into and we end up with things like enterprise class websites that cost millions of dollars to produce crumbling under the load of a dozen simultaneous users.
There are quite a few applications that do something similar, they leave a "disabled=1" or similar in the config to make sure people look at the config before trying to run the software.
I remember the eggdrop IRC bot doing it (<a href="http://cvs.eggheads.org/viewvc/eggdrop1.6/eggdrop.conf?view=co" rel="nofollow">http://cvs.eggheads.org/viewvc/eggdrop1.6/eggdrop.conf?view=...</a> , look for the lines starting with die) this and I'm sure there are more.
I recall reading that the "no brown M&M's" clause was added after a near-fatal accident on stage where a member of the Van Halen band got electrocuted because of bad wiring on the stage.
An episode of the TRC podcast covered this, and came to a different conclusion than snopes.<p><a href="http://www.trcpodcast.com/trc-219-can-men-and-women-be-friends-foods-with-animals-van-halen-mms/" rel="nofollow">http://www.trcpodcast.com/trc-219-can-men-and-women-be-frien...</a>
Once I buried a crazy request in a list of "you need to agree on these points or the book won't make the deadline" email to my publisher. My editor flat out agreed to them all.<p>That's how I knew she was lying about having read them and I had to escalate to the production editor.<p>It saved the book.
You can listen to Ira Glass and John Flansburgh of They Might Be Giants talk about it in the prologue: <a href="http://www.thisamericanlife.org/radio-archives/episode/386/fine-print?act=0" rel="nofollow">http://www.thisamericanlife.org/radio-archives/episode/386/f...</a><p>"there's 30 people the promoter's going to hire on our behalf ... but in only half of them did we require that they be sober"
I’ve heard similar stories regarding developers and IT Services<p>Where the devs weren’t allowed access to the Production environment so would have to leave written instructions on how to deploy the software they’ve written. And convinced that IT Services weren’t reading their instructions they would write something really offensive in there and see if they complained<p>Possibly just a myth, but amusing all the same
I've read thru a bunch of other nit-picky riders. Strikes me that an under-discussed factor is that these high-value stars (contracts running into the $millions) are under <i>extreme</i> pressure, which is severely aggravated by so much change on a daily/hourly basis; something as "trivial" as wrong-temperature or brand drinks (I dislike Poland Spring water, and prefer Mt Dew in cans not bottles), uncomfortable seats, or even brown M&Ms (hey, everyone has a pet peeve) can be an unnerving "last straw". Having a few "perfect" arrangements everywhere gives them something to center on for mental stability.<p>ETA: I realize this is a tangent. Methinks it's relevant.
<a href="https://www.youtube.com/watch?v=YwHO2HnwfnA" rel="nofollow">https://www.youtube.com/watch?v=YwHO2HnwfnA</a><p>Interview with Eddy Van Halen, telling the story first hand. :)
Interesting lesson about getting confidence by adding bugs. This was mentioned at a recent tech talk about Java Mutation Testing and PIT <a href="http://pitest.org/" rel="nofollow">http://pitest.org/</a> - video here <a href="http://vimeo.com/89083982" rel="nofollow">http://vimeo.com/89083982</a>
I'm stealing this tactic when interviewing a QA guy, if I ever do end up looking to hire QA guys, that is.<p>Me: "So here are a set of instructions are programmers were asked to follow. Can you see anything wrong"<p>QA candidate: "Why yes. They forgot to remove the brown M&M's"<p>Me: "You start tomorrow."
This reminds me of some IRCd configurations, in which the server will not function properly unless you've thoroughly read through the conf file and found the single commented line which disables the entire process.
Maybe a more interesting question is whether they ever exercised their right to terminate for brown M&M's.<p>Is there any notion of material breach, major vs minor breach, etc. in "tour contracts"?
I've seen same kind of tests on one of the fire ranges. You have to read safety rules. One of the points was to put x mark on the 2nd page if you read that.
Instantly thought of this:
<a href="https://www.youtube.com/watch?v=k_7kg5ZzDZo" rel="nofollow">https://www.youtube.com/watch?v=k_7kg5ZzDZo</a>
I'm not a lawyer, but as I understand contract law, delivering something "close enough" is all that's required to satisfy a contract. Let's pretend everything's perfect except for one brown M&M. I'm sure a lawyer can explain it better, but if everything else is in order, I think Van Halen would have to perform their end of the deal.
Maybe a mathematician can explain whether this "trick" works or not? Intuitively, I can't see that knowing whether the M&M demand was filled makes it more probable that the other demands are filled.<p>Say you have a pile with five black or white marbles. You want them all to be black. So you check that the first marble in the pile is black (ie no brown m&m:s). Is it now <i>more probable</i> that <i>the other four marbles</i> also are black?<p>Because you are just checking one specific marble instead of sampling a number of randomly chosen marbles (which of course would increase the probability), I don't see how it can work.