Science and Technology links (February 22nd 2020)

    1. In a large cohort study, the highest probability of reaching 90 years old was found for those drinking between 5g and 15 g of alcohol per day. This does not mean that if you are not drinking, you should start.
    2. The Earth is getting greener thanks to CO2. In turn, a greener Earth will mitigate global warming. (Source: Nature)
    3. In 2019, the carbon emissions in the US fell by 2.9%. They fell by 5% in the European Union. They also fell in Japan. (Source: Bailey)
    4. Robots can take blood samples and apparently do competent job, according to a clinical trial.
    5. We may soon get 80-terabyte disk drives.
    6. The age-adjusted cancer rate in the US is currently about at the same level as it was in 1930. We are not winning the war against cancer.
    7. You are better off cutting your food on wooden planks, they are more hygienic that plastic planks.
    8. Science is undergoing what some people call the “reproducibility crisis”: may important results reported in prestigious venues cannot be reproduced by other scientists, independently. Miyakawa suggests that the reproducibility crisis might be related to the fact that studies are frequently fabricated:

      (…) more than 97% of the 41 manuscripts did not present the raw data supporting their results when requested by an editor, suggesting a possibility that the raw data did not exist from the beginning.

      A few years ago, I was on the PhD committee of a student. I questioned the results. Ultimately, we asked for the software that produced that data. The student quickly reported that the software had been lost, deleted by the University. We declined to grant the PhD despite an extensive publication record (with articles in some of the best venues). I paid a political price for my choice to fail the student. The student eventually did get a PhD after an appeal. I would not be surprised to learn that this student became a professor. The lesson is that you should always doubt scientific studies. Ask that they be independently reproduced.

Daniel Lemire, "Science and Technology links (February 22nd 2020)," in Daniel Lemire's blog, February 22, 2020.

Published by

Daniel Lemire

A computer science professor at the University of Quebec (TELUQ).

6 thoughts on “Science and Technology links (February 22nd 2020)”

  1. If there is doubt on results published somewhere, you should at least note the doubts in for example. So that others can see there was doubt if they can’t reproduce the results.
    We’re now seeing post-publication peer review to become more and more important, and this seems to be the right place to use them.

  2. RE 2. the earth is getting greener

    Modelling indicates that greening could mitigate global warming by increasing the carbon sink on land and altering biogeophysical processes, mainly evaporative cooling.

    As much as I like and agree with this paper’s non-alarmist stance, I think any model that indicates an increased carbon sink without a new mechanism to precipitate carbon out of the system is flawed. Carbon emissions are outpacing the carbon cycle’s capacity to adapt. That fact is captured in atmospheric concentration measures. The answer to flawed models is not equally flawed models.

  3. RE 4. Robot Blood Samples
    The original paper says:

    The device demonstrated results comparable to or exceeding that of clinical standards, with a success rate of 87% on all participants (n = 31), a 97% success rate on nondifficult venous access participants (n = 25), and an average procedure time of 93 ± 30 s (n = 31). In the future, this device can be extended to other areas of vascular access such as IV catheterization, central venous access, dialysis, and arterial line placement.

    That is cool and especially interesting when compared to the Theranos 1.0 device (failed microfluidics) and 2.0 device (failed robotic automation). This device is the front-end to a Theranos 2.0 style system and removes the need for the specialized walk-in clinic partners that Theranos depended on. As a side note, it is a shame that the Theranos debacle has distracted us from the key lesson from the failure of the 1.0 microfluidic device: independent microfluidic test interact in unexpected ways when combined in close proximity to one another. Lab-on-a-Single-Strip is still achievable I think.

  4. RE 5. Heated Disk Drives

    I wonder why the energy requirements are not mentioned; it seems like a key constraint along with thermal characteristics.

  5. Since universities do not care about experiment data, and publishers care a lot and really do not want the data to be included with articles, what really did anybody expect?

    Do people really expect researchers to keep the data forever by themselves, with no help and nobody even caring about it?

    1. Some journals ask for the data and the software. I have published some papers where someone actually looked at my data and my software and vouched for reproducibility.

      I cannot speak for all fields, but there are certainly communities in computer science where making your data available is viewed positively.

      Overall, you are correct that the incentives are not aligned to favour data conservation.

Leave a Reply

Your email address will not be published.

You may subscribe to this blog by email.