It is the fashion in mainstream physics today to always start from the existing theory. For example, general relativity is always assumed to be right. If you don't believe that, try questioning it and see what happens! As a result the mainstream need to work out what data they need to find to make it right. Hence the search for dark matter, dark energy, dark flows, which brings in lots of funding too. This is the process everywhere, but it is the opposite of the scientific method which puts data first and reigned between say 1660 (founding of the Royal Society, who said 'disregard theory and look at data') and 1988 (when data-driven Feynman died). If you want to be cheeky, you could call the post-1988 way the 'religious' method, but without the attached morality.
Probably because I was educated in a more grounded form of physics (BSc in physics, PhD in ocean physics) and loved reading Feynman, I am pre-1988. What I like to do, and have ever since my physics degree, is look for interesting anomalies (data that defies the theory). Actually, before my physics degree I was fond of theories and philosophy and did not bother much about data. I spent hours in the library reading about Spinoza, and trying to devise theories from beautiful thoughts alone, but something changed when I did my third year research project at York University: An analysis of a chaotic Duffing oscillator. I built such an oscillator in the university's metalworking lab. It was a beautiful thing and I wish I still had it! (see my schematic below). A metal pendulum with a magnet at its base, repelled from its equilibrium point by a magnet underneath. It had two side-arms with magnets attached pointing down. One arm was driven sinusoidally with a electromagnetic coil around the magnet, the motion of the magnet on the other side was sensed with another electromagnetic coil. The signal was fed to a BBC computer, that also by integration could work out from the measured speed, the position of the pendulum. I collected and plotted strange attractors of the chaotic motion - the pendulum oscillates between two stable positions chaotically.
When I started my PhD shortly after, I began reading Feynman's books. Also, I eventually focussed on a beautiful anomaly. Cruise data has shown that every summer, a thin cold, fresh surface layer spreads over the north Atlantic. Why? I built a simple layered computer model of it, showed the spreading was due to wind-driven (Ekman) flow blowing polar water south and showed that the air-sea interaction heated the cold surface as it went, but did not erase the freshness, so it becomes unexpectedly buoyant (being now warm and fresh, both properties reduce water density). It forms an insulating cap on the ocean that has implications for climate (paper).
Later when I worked at the Met Office I was tasked with looking at the output of the ocean model and I decided, being fond of data by now, to look at the output without the smooth interpolation that was being done. I pixelated the raw sea surface salinity data instead, and what immediately appeared were nice bands of fresh surface water underneath rainbands. So I developed a simple model of those as well, and that predicted consequences for weather too.
I've always been keen on fundamental physics & astronomy and so I couldn't help but notice anomalies like the galaxy rotation problem, the Pioneer anomaly and that they both involve the same odd acceleration 10^-10 m/s^2. I developed a simple model to explain those, called MiHsC or quantised inertia and it turns out it predicts a lot of other anomalies, such as the emdrive, and cosmic acceleration which I did not know existed till it heard about it on the car radio and thought "MiHsC predicts that!"
I do love looking for anomalies or mysteries. That is why mainstream physics now seems so dry because they are so confident that they know it all and anomalies are brushed under the carpet with arbitrary fudges like dark matter. In my latest attempt to fight back, I have started writing #AnomalyoftheDay on twitter, documenting all the well-observed anomalies that prove that physics is very incomplete (eg: it only predicts 4% of the cosmos). There are many anomalies now, from the proton radius being different depending on how you measure it, the gravitational constant not being constant (blog), tapered microwave ovens which thrust slightly without expelling propellent (emdrives), odd lights flying around in Hesdallen, Norway (link), galaxies rotating in violation of Einstein, and the Cosmic Microwave Background being aligned with the Solar system in a way that would make Copernicus weep! (paper, see Figs. 1 and 2). I have a list of 40 or so anomalies and it is growing.
The tendency I and some others are fighting in mainstream physics is a huge one, a combination of hero-worship, intellectual laziness, group-think and a bias in physics towards mega-expensive solutions like dark matter detectors since bringing in the most funding gets academics promotion. My hope is to get physicists to look up from old books and funding applications and look at real anomalies again (an act which requires little or no funding and repays you with fun), or at least get taxpayers to demand they do. Only then will the mainstream see the utility of quantised inertia.