Navigation

BOGUS RESEARCH: The best programmers are 10x better than average programmers.

“In the study, two groups of six subjects each, comprising a total sample of 12 experienced programers, debugged two types of programs under online and offline conditions in accordance with a Latin-Square experimental design.”

This result has been directly cited 244 times. The most popular corruption is from Brooks 1987 paper, “No silver bullet – Essence and accidents of software engineering” (which itself has received 3645 citations):

“Study after study shows that the very best designers produce structures that are faster, smaller, simpler, cleaner, and produced with less effort (Sackman 1967). The differences between the great and the average approach an order of magnitude.”

The standard citation for the claim that “the best programmers are 10x better than average ones” actually says something terrifyingly different:

“The horrid portion of the performance frequency distribution is the long tail […] in which one poor performer can consume as much time or cost as 5, 10, or 20 good ones.”

Translation: A small number of the very worst programmers are very bad. These bad programmers are 5-20x worse than the very best. However, the best programmers are only 2-3x better than “average” ones in this study.

Furthermore, nearly all the statistical variation in performance is accounted for by literally one single programmer in a 46 year old study (n=12). This study was actually conducted to see if giving people access to computers that ran their programs immediately instead of waiting days for their code to be run on shared systems improved performance (News flash: It did).

This result has been directly cited 244 times. The most popular corruption is from Brooks 1987 paper, “No silver bullet – Essence and accidents of software engineering” (which itself has received 3645 citations):

“Study after study shows that the very best designers produce structures that are faster, smaller, simpler, cleaner, and produced with less effort (Sackman 1967). The differences between the great and the average approach an order of magnitude.”

See on www.docstoc.com

Author Description

Louie Helm is a Machine Learning Engineer

Recent

  • More People Signing Up For Cryonics

    March 4
  • Asparagus Metabolizes Into Neurotoxin Over 5x More Powerful Than PCP

    June 18
  • Cryonics Scientifically Proven To Work In Model Organism

    May 23
  • Longevity Cookbook Indiegogo Launch Party

    May 4

Popular

  • Darkcoin Rising

    May 14
  • Increase Longevity And Intelligence With Boosted Klotho Levels

    July 25
  • Chore Reduction Device Cleans Your Clothes In 10 Minutes

    November 5
  • Betterment vs. Wealthfront vs. LendingClub

    June 23