this will probably void your warranty.

"In other words, there is no reason to believe that the trivial variations in mortality risk observed across an enormous weight range actually have anything to do with weight or that intentional weight gain or loss would affect that risk in a predictable way."

"

But here’s the thing: when the economists were shown both the graph and the detailed numbers, the number of economists getting the answer spectacularly wrong — the number giving an answer of less than 10 — soared. Just working with their eyeballs, 3% of economists got it wrong. Working with the numbers as well, that proportion rose to 61%! And when a third group was given the numbers and no chart at all, fully 72% of them — professional economists all — got the answer badly wrong.

…

I’m certainly guilty of this kind of thing: I see a paper demonstrating a statistically significant correlation between one variable and another, and I generally assume that if the experiment were repeated, we’d see the same thing again. But that’s not actually true.

And so it’s easy to see, I think, how economists become convinced of things that the rest of us aren’t sure of at all — and how the economists often end up being wrong, while the rest of us were right to be dubious.

What’s more, if economists are bad at this kind of thing, just imagine what other social scientists are like, or even doctors. Next time you see a piece of pop-science talking about interesting findings from some paper or other, bear this in mind. A lot of papers are written; a few of them have interesting findings. Those are the papers which tend to get publicity. But there’s also a very good chance that they don’t actually show what the headlines say that they show.

"

"Thomas Jefferson and George Washington recorded daily weather observations, but they didn’t record them hourly or by the minute. Not only did they have other things to do, such data didn’t seem useful. Even after the invention of the telegraph enabled the centralization of weather data, the 150 volunteers who received weather instruments from the Smithsonian Institution in 1849 still reported only once a day. Now there is a literally immeasurable, continuous stream of climate data from satellites circling the earth, buoys bobbing in the ocean, and Wi-Fi-enabled sensors in the rain forest. We are measuring temperatures, rainfall, wind speeds, C02 levels, and pressure pulses of solar wind. All this data and much, much more became worth recording once we could record it, once we could process it with computers, and once we could connect the data streams and the data processors with a network. How will we ever make sense of scientific topics that are too big to know? The short answer: by transforming what it means to know something scientifically."

"Eventually consistent semantics provide almost no guarantees regarding the recency of data returned (unbounded staleness of versions). Despite these weak guarantees, many data store users opt for eventual consistency in practice—why? It’s often faster to contact fewer replicas, and it’s also more available. However, instead of relying on anecdotal evidence, we should quantify why eventual consistency is “good enough” for many users. We can predict the expected consistency of an eventually consistent data store using models we’ve developed, called Probabilistically Bounded Staleness. It turns out that, in practice, and in the average case, eventually consistent data stores often deliver consistent data. Using PBS predictions, we can optimize the trade-off between latency and consistency and better understand why so many data store users choose eventual consistency."

"It’s my view that if you put the best scientists, science communicators, and science journalists in a room, it wouldn’t take long for them to agree on the basics of good medical science reporting.

A checklist would look something like the following. Every story on new research should include the sample size and highlight where it may be too small to draw general conclusions. Any increase in risk should be reported in absolute terms as well as percentages: For example, a “50 percent increase” in risk or a “doubling” of risk could merely mean an increase from 1 in 1,000 to 1.5 or 2 in 1,000. A story about medical research should provide a realistic time frame for the work’s translation into a treatment or cure. It should emphasize what stage findings are at: If it is a small study in mice, it is just the beginning; if it’s a huge clinical trial involving thousands of people, it is more significant. Stories about shocking findings should include the wider context: The first study to find something unusual is inevitably very preliminary; the 50th study to show the same thing may be justifiably alarming. Articles should mention where the story has come from: a conference lecture, an interview with a scientist, or a study in a peer-reviewed journal, for example."