Are We Already At The "End Of Work"?

Tyler Durden's picture

Submitted by Charles Hugh-Smith of OfTwoMinds blog,

The Python That Ate Your Job

We are already well into the "end of work."

The more accurate title would be "The Python (Script) That Ate Your Job." Python is a computer language whose core philosophy is summarized by "PEP 20 (The Zen of Python)", which includes aphorisms such as:

  • Beautiful is better than ugly.
  • Explicit is better than implicit.
  • Simple is better than complex.
  • Complex is better than complicated.
  • Readability counts.

(source: Wikipedia)

As I understand it (from a non-programmer POV), Python enables rapid development of scripts that may not be optimized by some metrics but which work perfectly well in terms of solving a problem in a cost-effective manner.

(Programmers can be highly partisan, i.e. emotionally attached to their preferred language, so I am trying to be as non-partisan and careful as possible here to avoid arousing the ire of either Pythoneers or Python detractors. I am just an ignorant bystander; please don't shoot the piano player, etc.)

A senior manager at a small tech company recently related a story that illustrates 1) the power of Python (and other scripting languages) and 2) the changing nature of work:

The company had some time-consuming data analysis that needed to get done on a regular basis, and the manager was considering recruiting a (paid) intern to do the work. Instead, he spent four hours writing a Python script which did the work in a few minutes. He named the program "Intern."

This story is repeated thousands of times a day across millions of tasks. Virtually all of my self-employed friends use technology to enable one person to produce output that would have taken three people in the 1980s.

As management guru Peter Drucker noted, enterprises don't have profits, they only have expenses. If you are self-employed or own/manage a business, you will immediately grasp the profound truth of this insight.

If you can replace an expensive worker (and every employee is expensive nowadays, due to the high cost of labor and general overhead) with a Python script that can be crafted in a few hours, financial fact compels you to do so: your business has no profit, it only has expenses.

This dynamic is scale-invariant, meaning it is true of all organizations, from one-person businesses up to global corporations and entire nations. A non-profit group only has expenses, and so do churches, cities and nations. Once expenses exceed income, the organization goes bust.

Could I be replaced with a Python script? In some ways, yes: a script could be written that mined the thousands of entries and essays I've written for repeating words, phrases and themes, and the script would rehash the material into "new" entries.

But since the script isn't logging "experience" in the same way as a human does, the script would not be able to replicate dynamics such as changing one's mind or taking a new direction, although it could randomly generate such behaviors to mimic human development.

Would the script be "good enough" to attract readers? Perhaps; but attracting and keeping readers is not necessarily a problem-state that can be solved with data-mining and pattern matching, as readers seek not just novelty and expressive writing but insight. Any script that rehashed existing material would not be generating new insight; it would simply be repackaging previous insights.

For highly partisan blogs, this might well be "good enough," since partisan readers actually want to read the same rehashed material again and again: in effect, a script that repackaged "it's the Demopublican's fault" with new headlines and slightly different content would closely match the human content generator's output.

I have no doubt some clever programmers have already played around with generating rehashed content and posting it as a blog written by a human being, an artifice masked by an avatar ("Hi, my name is J.Q. Public and I write about politics."). It would almost amount to sport to generate a phony history and cobbled-together quirks to fill out the illusion of personhood.

(Some readers have even wondered if "Charles Hugh Smith" is such an avatar. The answer is no, because the history and quirks of "Charles Hugh Smith" are simply too implausible to be believable. Also, the cost of maintaining such a complicated avatar isn't worth the paltry income generated by the blog. What machine intelligence would be dumb enough to maintain this idiotically complicated enterprise for such a paltry return? Only a human would be compelled to do so.)

Could a robot and standardized scripts replace everything I can do with a Skil 77 wormdrive power saw? It could certainly do a great many repetitive tasks at a work bench, but it would not be able to do non-standardized, on-the-jobsite tasks such as cutting out the rotten sections of a wood window frame. The robot might be able to execute the cuts (presuming it was light enough and mobile enough to stand securely on a scaffold or slope), but it would need a human partner to program the cuts in the real world and in real time.

In other words, "work" is increasingly a partnership of humans and technology. If one's skills and experience (i.e. labor) can be replaced with a Python script, it will be replaced by a Python script. Organizations that fail to replace costly paid human labor with a script will have much higher costs than those organizations that replace paid labor with scripts.

The paid human labor that can't be replaced by a script will increasingly require the knowledge and skills needed to collaborate with technology as an essential work partner.

We are already well into the "end of work." Digital pythons have been eating jobs for some time now, and because organizations only have expenses, they will continue to do so indefinitely until the only paid jobs left are those that cannot be fully replaced by a script or a robot operating on standardized scripts.