Actuarial work, bioinformatics, social science. These are data-driven cultures. They actually require scientifically valid and methodologically sound data to make a claim.
Programming, on the other hand, is dominated by fashion. Language wars, methodology wars, business bullshit, buzzword chasing. Programming is not data-driven AT ALL! When was the last time a computer scientist actually did some SCIENCE? When was the last time a programmer actually ran a double blind study?
99% of programming is not data-driven whatsoever. The sweeping decisions in programming are made by corporate big wigs operating on their intuition, or are design choice (extremely subjective!) made by "architects" who, for instance, created UNIX.
Were the people who created Python, Ruby, Java, C, etc DATA-DRIVEN? What studies did they use to decide that so-and-so feature should be like this and not like this?
Programming is mostly a craft and has essentially nothing to do with being data-driven. Doing A-B tests does not mean your culture is data driven when A-B tests are like 0.1% of everything you do. And most A-B tests are methodologically unsound anyway and would be shamed out of any real social science department.
Economics, on the other hand, is an actual science with actual data that performs actual methodologically sound studies using advanced statistics. Practicing economists have to use actual valid data procured from real studies to have careers. Programmers mostly twiddle their bits around until something works. That IS NOT being data driven.
One of the most maligned professions when it comes to forecasts is meteorology. Nate Silver does and excellent job here, shedding some light on the challenges and triumphs of predicting the weather.[1]
Software engineering can involve difficult models about different scaling scenarios. Civil engineering might involve unexpected surprises about how standing waves emerge in bridge design. Financial analysis involves a forecast of the total revenue stream the company will generate between now and the end of time, and a guess about what other market players will predict about the future a quarter from now, since the stock price, too, can affect its income.
Take P/E ratios for example. Should you look at them as a sanity check, or think of them as a broad measure of the market's beliefs about the issue's future growth potential?
Regardless of where you stand, it is patently absurd to state that financial analysis is not data driven, and the decision making does not reward empirically successful results. Whether you recommend your fund bets with or against the market you get less of a say next time when you have less money left to bet, or decrease your assets under management by losing your clients' money.
Programmers are not immune to that failure mode, but programming culture encourages data-driven decision-making and prizes empirical results.