Bad Computer Code Jeopardizes Results of 100 Academic Studies – IOTW Report

Bad Computer Code Jeopardizes Results of 100 Academic Studies

Breitbart:

Scientists at the University of Hawaii have uncovered a computer glitch that brings into question the findings of over 100 academic research papers.

According to a report by Vice, scientists at the University of Hawaii discovered that a commonly-cited chemistry research paper was built around faulty computer programming. The original study’s authors issued an update to their paper last week, explaining that 1,000 lines of code that helped determine the outcome of the study were faulty.

The researchers claim that faculty coding was the product of a glitch that results from code running on multiple operating platforms.

An assistant chemistry professor at the University of Hawaii at Mānoa utilized a piece of code that was improperly programmed. The code was used to determine the molecular makeup of samples that are studied in laboratories. read more

8 Comments on Bad Computer Code Jeopardizes Results of 100 Academic Studies

  1. Programmers think they know everything, but there are many things they don’t know — and it sometimes creeps into their code.

    The date in Microsoft Excel is, in theory, the number of days since the beginning of 1900, with January 1, 1900 as day one. It includes February 29, 1900. Unfortunately, 1900 was not a leap year. That may not be a big deal, but I have seen applications that let you look up dates going back as far as the mid 1700’s and as far ahead as the mid 2200’s, including February 29 in the years 1800, 1900, 2100 and 2200. None of those years are leap years

    But don’t try telling that to the programmers, because they know oh-so-much more than you.

    The big difference between the Julian calendar and the Gregorian calendar is that EVERY four years was a leap year in the Julian — that’s why it got so far off. The Gregorian calendar has a leap year every four years, unless the year is divisible by 100 in which case it isn’t, unless it is divisible by 400 in which case it is (and is why 2000 was a leap year).

    6
  2. Oh boy, the university just published another correction. It isn’t a bug about measuring molecular makeup, turns out climate science has been built on this bad code…. U of H added it looks like the kooks from 1972 who predicted a new ice age were right.

    8
  3. @Matthew Meaney October 16, 2019 at 1:57 am

    > Programmers think they know everything, but there are many things they don’t know

    There is an entire village of United Statesians that have, now, sworn vengeance on your family’s tech support for seven generations.

    2
  4. Zonga OCTOBER 16, 2019 AT 2:50 AM

    U of H added it looks like the kooks from 1972 who predicted a new ice age were right.

    Well yeah. It’s gotta be something.

    You can’t just live life without fear and have modern conveniences like food and cheap energy when you need to be controlled and made miserable!

    7
  5. Ok… This one is kind of in my lane. So…

    1. The code is written in Python. This is a portable scripting language. It allows you to write this once, and run it on different operating systems with usually very little modification, provided you don’t access system specific functions. Generally speaking most CS types consider it a language for quick hacking. But it’s been pushed into areas where people used to use languages with more formally specified floating point (FP) math abilities, like C/C++, Java, Fortran, etc… And I only include Java begrudgingly here. Its had significant FP problems too, but it’s used by more CS types that correct for this.

    2. The code was written by a scientist, not a trained programmer. An actual computer science major would have learned about FP code portability in Freshman CS 101.

    3. This is how science is actually supposed to work. Errors get detected and corrected. (Que: climate change jokes…)

    The actual paper is here:
    https://pubs.acs.org/doi/10.1021/acs.orglett.9b03216

    1

Comments are closed.