Play and hypothesis testing: From a note to my research group

I do not believe in the model that science advances by testing hypotheses. I think that people play and then stumble across something interesting and then later repose what they stumbled across as a hypothesis to be tested.

I think we need to do two things simultaneously:

  1. We need to have fun and be playful and work on things that seem interesting even if there is no obvious high impact paper to come out of it.
  1. We need to focus on writing high impact papers that make important novel contributions.

Most scientists essentially rewrite the same paper over and over again throughout there careers, as they slowly drift off into irrelevancy.  This is what happens when people just go on just treating aspects of problems not treated in their previous papers. To remain vital, we must address new problems.

So, the questions are:

A. What are avenues of play likely to be fruitful in leading to interesting, relatively unexplored territory?

B. What are potential important high impact papers that we can be working on now?

Reversal of radiocarbon flux into the ocean

Radiocarbon is produced naturally in the stratosphere when cosmic rays strike atmospheric nitrogen. Most of it decays in the deep ocean. The natural pattern is for radiocarbon to be produced in the stratosphere and mix through the atmosphere. It is then absorbed by the ocean where it decays. So, the net flow throughout most of this planet’s history is for radiocarbon to go from atmosphere to ocean.

This natural pattern is disturbed by two different sorts of human activities — both disturbing in other ways.

First, humans tested many nuclear weapons in the atmosphere mostly in the 1960’s and this nuclear weapons test more-or-less doubled the amount of radiocarbon in the atmosphere.

Second, humans have been burning a lot of fossil fuel. This ancient carbon is devoid of radiocarbon.

As the CO2 emitted by humans acidifies the ocean, this acidification tends to drive radiocarbon out of the ocean.

Thus, nuclear weapons tests tend to drive more radiocarbon into the ocean and fossil-fuel emissions tends to drive radiocarbon out of the ocean.

In a 1998 paper (Caldeira et al., 1998;, we wrote “we, predict that atmospheic 14C content will reach a minimum and start to increase within the next few years if fossil fuel burning continues according to a ‘business-as-usual’ even though fossil fuels are devoid of 14C.”  That is, we predicted that the effect of fossil fuel burning would start to overwhelm the effect of the nuclear weapons tests.

If you look carefully at a figure in the supplemental material to a recent paper published in PNAS (Graven, 2015;, in supplementary Figure S1 panel a, you can see that atmospheric radiocarbon content indeed started increasing at around that time, suggesting that our understanding of global carbon cycle functioning is largely accurate.

This is happening because stratospheric radiocarbon production is now exceeding the rate of ocean uptake of atmospheric radiocarbon.

From the perspective of confirmation of model predictions, this is heartwarming to see. It is nice to make a model prediction and see it later confirmed by observational data. From the perspective of the Earth system, it is of course a disturbing result.

In her recent paper, Graven emphasizes how adding all of this ancient carbon to the atmosphere is making the atmosphere look old from a radiocarbon perspective, which will have obvious implications for future efforts to use radiocarbon to date samples. Plants grow by getting their carbon from the atmosphere. If we add a lot of ancient carbon to the atmosphere, we make modern wood look ancient.

I find it remarkable that, as we pointed out in our 1998 paper, sometime around mid-century, ocean acidification from our fossil-fuel CO2 will cause radiocarbon to stream out of the ocean into the atmosphere, reversing the natural direction of flow.

For billions of years, the radiocarbon flow has been going from atmosphere to ocean. There is no time known in Earth history when this flow has been reversed. If we continue current trends in fossil fuel combustion, sometime around mid century, this flow will reverse and radiocarbon will stream out of the ocean into the atmosphere. This may not have huge practical implications outside of radiocarbon dating and similar pursuits, but this reversal acts as a reminder of the magnitude of humanity’s impact on the environment, and how geologically unusual our modern era is.

It is great to see Heather Graven extend our work and show how important fossil fuel burning will be for future efforts to use radiocarbon to estimate the age of archaeological materials.  Our earlier paper was more philosophical and less practical, and was largely ignored. (It has only been cited 9 times since being published 17 years ago.) I thought the fact that we will be reversing the direction of radiocarbon flow that has persisted for billions of years would be of broad interest and receive widespread attention. It did not. I am grateful to Heather Graven for giving our work some attention and extending it in new directions.

Do open discussion journals provide a service to the scientific community?

The open “discussion” journals favored by the European Geophysical Union (EGU), including Atmospheric Chemistry and Physics Discussions (ACPD), may be doing the scientific community a disservice.

For those not initiated into the ways of these journals, scientists submit manuscripts to Atmospheric Chemistry and Physics Discussions where they are published without peer review, and then later if successfully peer reviewed, a modified version appears in the regular journal Atmospheric Chemistry and Physics (ACP).

This produces at least two versions of every paper: a non-peer-reviewed draft and and a final peer-reviewed version. The continued circulation of the non-peer-reviewed draft can act as a kind of pollution of the scientific literature, as it is often unclear to the uninitiated that papers published in EGU “Discussions” journals are not peer reviewed.

One of the key contributions of the editorial and peer-review process provided by journals is the vetting of the scientific content both for importance and quality. There is way too much stuff being written to read everything, and the editorial process at a high-quality journal is supposed to help provide a filter and direct scientists to important, high-quality, papers.

It is a fiction to believe that busy scientists have the time to review a panoply of manuscripts that they are not specifically tasked with reviewing. Journal editors know how hard it often is to obtain thorough reviews of submitted manuscripts.

There may be other models that would allow volunteer reviewing without expanding the reservoir of grey literature. For example, journals could list titles of papers sent out to review, and volunteer would-be reviewers could contact the journal and ask to be made a reviewer.

By publishing papers that are not peer reviewed, EGU journals such as ACPD are contributing to the noise of science, when the role of the editorial process should be to help readers find the rare nuggets of important high quality signal amid the abundance of excess noise.

[Note: This post was prompted by Andy Revkin’s piece of DotEarth: ]