While computational models have made big strides there is a large gap between model predictions and the real world. If you are serious about how something interacts with the air moving by it you need access to a wind tunnel. This gets interesting with athletics. At the highest levels in cycling wind tunnels are used, but only sparingly as wind tunnel time is hard to come by. It was interesting that one of the big three bicycle makes recently opened a private wind tunnel to sort out bicycle aerodynamics.
I'd much rather see focus on the real issues of increasing bicycle commuting, but the fact is that much of the upscale market in the US is built on the bike as an expensive toy and exercise machine. Competition in making racing bikes is fierce and I expect the other major makes - especially Trek - to build their own facilities in the near future.
Of course I'm assuming they will have people who know how to properly design the experiments and analyze the information. It makes me think about the current big data explosion. Some will do great things, but getting there isn't easy and requires very careful design and the right people. Just going out and hiring "data scientists" and "data vis" people won't work and may even be counterproductive.
Any time you are dealing with data (I would say information) analysis, it is important to know what it is, where it came from, how noisy it is, and possibly many other things. You have to know about the manipulations and filters it passes through. Are your questions for the data set? How biased is your analysis?
Look up at the night sky and constellations pop out. Constellations are artificial and offer no physical meaning. In fact they led to early counterproductive models of the cosmos and some people still attach meaning to them.
Humans are naturals at apophenia - we have evolved to easily find artificial patterns in information. We find it in scientific, engineering, financial, weather and most other forms of data. We can and do make serious mistakes. Hunting for patterns in large data sets can be enormous fun, but it can be very misleading - as the Danes would say: there are owls in the marsh.... A deep tool and domain knowledge is essential. Given how many misuse spreadsheets and simple statistics, I'm not universally optimistic. But others will be very good at avoiding the pitfalls.
There are ways to guard against mistakes. Science is built on these procedures, but there are those who see big data as the shiny new thing.
I was interested in measuring how efficient a human could be on an average bike. There are a lot of numbers, but they vary dramatically. Using a numeric model and a digital wind tunnel we are talking about serious big data. For fun one can get time in a low speed wind tunnel. There is much more going on than in the digital model, but a bit of clever design means we are not recording that many numbers. With some care we can understand what is going on throughout the experiment - something that isn't as easy with the digital model.
The opportunity presented itself and I was able to record a nice round number - From an earlier post a friend on a commuter style bike can exceed the equivalent of 1,000 miles per gallon. a few details:
Looking at someone in good physical condition riding in an upright position with no external wind on a very non-aerodynamic bike. At 22.5 kilometers per hour (14 mph) she needs about 80.6 kilojoules to cruise along for a kilometer. This works out to about 31 nutritional calories per mile. She is burning something like 434 calories over her basic needs to cycle for an hour.
It is interesting to calculate the power she is getting from the metabolism of her food at this steady pace. In an hour she requires about 1814 kilojoules of energy, so dividing by the seconds in an hour we get a bit more than 500 watts. Not all this food energy is being converted into useful mechanical work. It turns out many of the muscle movements we make have efficiencies around twenty percent. Colleen happens to be a trained athlete with wonderfully smooth motions and is a bit over twenty percent efficient on a bike. She is delivering about 100 watts of power to the pedals and most of that (about 95%) is making it to the rear tire.
Imagine a car that gets 30 miles per gallon. A bit of arithmetic shows 30 mpg of gasoline is about 2750 kilojoules per kilometer. The car uses a bit more than 34 times the energy Colleen does to travel the same distance. Of course she is only carrying around a bike and a car has to carry a lot of weight besides the driver.
She is close to 1025 miles per gallon - if fueled by a gasoline near equivalent like a vegetable oil. That doesn’t work for her so you can figure out what she gets on a gallon of Ben & Jerry’s if you like.
There are a lot of ways to come to different numbers and not everyone has the access to a wind tunnel and the metabolic measurements that an Olympic program uses. If the information is found it is necessary to know something about it. There are a lot of specialized conditions that go into this little number. One can ask many other interesting questions and learn even more, but I only show the rather flashy economy number.
In the end just doing it was much more efficient that building a model and using computational fluid dynamics. If I was asking a more isolated question - perhaps a design question - the model may have been the way to go. A bit of experience goes a long way.
My worry is that many decisions come down to needing a simple and clearly stated result. The culture of those who describe the information and those who act on it may be so different that the analysis is worthless or even dangerous. Domain knowledge is critical. Even Nate Silver gets silly when he is away from politics and sports.
You have to be careful and critical - doubt and curiosity are very important tools as is playfulness. Science is lucky. There is an absolute right that can be independently tested using multiple approaches. Along the way there are many wrong descriptions, but over time they are stripped away and a clearer picture emerges
There isn't any magic in it. Big data will be great for some organizations, and others will get results that lead them down mistaken paths (ask Mitt Romney). How an organization builds this into their culture is critical. It should also be stressed that some organizations that have a good understanding of their domain using conventional techniques may find this counterproductive. Other companies may find it much more useful to continue to build more conventional techniques to understand their customers and operations - Trader Joe's comes to mind.
I've been seriously occupied for the past few weeks and did very little in the way of interesting cooking. So here is a fundamental technique I use all the time - brazing and glazing vegetables. It is very useful to master this one:-)
° Cut some firm vegetables (I love this with sweet potatoes, it doesn't work well with produce that isn't firm) into roughly equal sized pieces so they cook at similar rates. Put them in a pan with a lid that fits well and is just big enough to hold them in a single layer.
° Add enough water to cover the bottom of the pan along with a little olive oil or butter. Add any seasonings that need cooking - garlic or shallots for example.
° Place the pan over a medium heat, cover and stir every few minutes to see if the veggies are getting tender. You can add a bit of water if needed.
° When they are getting tender, remove the cover and crank up the heat to high. Stir constantly and wait for the water to evaporate and the veggies to start to turn brown - this many only take a couple of minutes.
° Add whatever spices you need. Salt, pepper, some chopped herbs etc. It often helps to add just a bit of acid - a squeeze of a lemon or a tbl of vinegar.