You're probably thinking of the Chung-on-a-stick
, though I believe Zipp has indicated they've used something similar mounted to riders' bikes to get a wind profile for some courses and optimize their rim shapes around those apparent wind angles.
For all of this "may not be applicable in the real world" talk, you guys are aware that field testing is a thing and that it does show correlation with wind tunnel testing right? http://www.cyclingnews.com/features/how-aero-is-aero
You can even do it yourself with a power meter!
Nobody is saying that you'll cruise around with a static 30W savings at all times, that would be ridiculous, but there are times where you will have a substantial gain from improved aerodynamics from a frame, wheels, kit, etc. like out on a break away, if you're taking a turn pulling into the wind, etc.
Of course there are tradeoffs, and those have to be weighed for each race, but the bikes absolutely have a place in the quiver of a rider's available tools to help them. And so what if there's only a marginal gain at times? The light 6.8kg bike would make even less of a difference in a flat 200m sprint than something even moderately more aerodynamic. Up a hill, protected in a pack, or where comfort may play a bigger role? don't bring it out then. They may not be the right choice for those of us who can only have one bike (like myself), but if you have easy access to it, why wouldn't you consider it as an option for those days when the race/ride plan will have use for it.
thing I do like about slowtwitch is that they're willing to say "show me the data" and consider the conclusions that come from it. Here it's "show me the data; no not that data, it conflicts with my strongly held, but ultimately baseless beliefs."