---
date: 2019-02-12
modified_at: 2019-04-25
tags: [entrepreneurship, lifestyle]
description: Exploring how predictive programming using neural networks could revolutionize software development and potentially reduce the demand for traditional programmers.
---
# Deep Dream Devs

- I find myself dreaming that I'm programming
- My brain is training and improving its neural network
- I think predictive programming CNN (PPCNN) could be the future of
  programming.
- PPCDN could map a textual description of what you want to a set of
  suggestions of implementations, which the programmer can quickly test and
  review.
- The PPNDN then improves with the feedback of the programmers choice
- This whole concept may completely change the way we develop software and
  greatly speed it up.
- When this happens, it could make the bubble of software developers demand
  diminish quickly since this way of programming could be much faster and
  easier.

## Can Microsoft do it?

Microsoft just bought GitHub. Their goal to be more Open Source is a perfect
strategy to make this look like they're doing it for the greater good. But what
if they're planning to change the way we do programming?

They already ask people permission to share insights into their private repo's
in settings.

All they need to do is to get a commit message as clear as possible, and be of a
single fix, so they don't have to know which change is which description. I bet
there is already a portion of professional programmers that does this very
strictly and clearly. They just need to find those programmers, and find the
projects (and commits) that do this well. Then, a CNN can be trained on this.
With millions of projects in GitHub, I bet there's more than enough data.

## The bubble is about to burst

When these kind of technologies become performant, and it actually becomes
easier to write code this way rather than normal programming.. Programming
things will become easier and progress in this will be fast. It is not uncommon
for CNN's to improve in accuracy (decrease their mistakes) by dozens of
percentages every year. This rate of improvement can't be done with regular
unassisted human programming.

When this happens, we will be able to achieve the same amount of work half the
time every few years. This means, we will do more, with less developers. The
need for programmers will decrease rapidly. Also, the skill of programming won't
be in the syntax, it will just be in the thinking of architecture and design.
Anyone who can think and can understand the needs of a project, will be able to
program. When this happens, the P.O. will be the best programmer.