Powerful Programming Languages, Take Two | Hacker News Reader

Powerful Programming Languages, Take Two

1 pointDanielBMarkham4 comments

So let's say it's 50 years into the future. You and your startup team get together for a weekend to crank out the next killer app. (Surely you haven't given up startups by now, have you?)

You all put on your neural helmets and begin to have a conversation about where the users are, what strategies you'd like to use, and the types of things you'd like to do in the future. At the end of this discussion, the computer creates a 10-minute neural "video" of your agreement along with a working application. Any newcomer to the team can "watch" the neural video and not only understand the application, but also understand what decisions were made and why, and how things should change in the future. The working application consists of 40K lines of assembly, automatically created during the discussion. To change the program, you just need to repeat the discussion process you did to create it -- ie, no assembler language knowledge is needed. Of course you _could_ go in and hack the assembler if you chose.

Conversely, at the same time, Arc++, The Next Generation is out. For this same amount of work, the code needed is this:

Floog()

Which of these is the more powerful programming language? Why?

If you believe in the great nano-hardware apotheosis, 50 years from now you won't program, you'll advise. Startup hackers will write essays like pg does now, except their audience will be swarms of semi-cognizant nanobots who will decide on their own if their suggestions are worthwhile. Only the efficient ones will have their choices passed along to the next generation. There will be great religious wars at the atomic scale as the bots try and decide where best to put their molecular parentheses. All this will happen so fast it will be unnoticeable to humans, most of whom will be too busy keeping each other entertained.

My crystal ball doesn't work much past the Great Singularity, so can we limit the discussion to somewhere half-way between here and there? At the extremes, a lot of discussion falls apart. Plus, I don't want no nano-bots arguing over parentheses. Little buggers should be using braces instead. : )

Long before the advent of computers, mathematical notation provided a way to more precisely express ideas than natural language. Of course, if you are verbose enough, you can express mathematical ideas with prose. But the corresponding mathematical notation will probably be more clear and easier to manipulate than the prose version for someone well versed in both "languages."

Programming languages can similarly provide for precise, concise expression of ideas.

Your example is a huge leaky abstraction waiting to happen. What if something about the resulting program is very different than what you expected or wanted? How do you know at what point things went wrong? How do you fix it? What if there were ramifications that were not clear at the high level of your conversation, but the machine dutifully did what you "wanted", with repercussions like Mickey in the Sorcerer's Apprentice?

The only way programming languages will no longer be useful in the future is if algorithmic thinking is no longer useful. Is that your assumption here?

Programming languages can indeed provide for precise, concise expression of ideas, but how precise must the description of a problem be in order for the computational machinery to express it within parameters necessary to satisfy the user?

So what if the program is very different that what you expected or wanted? Is this such a new thing? A better question would be "what if I could be wrong eleven times in a week and finally be happy. Would that be better than taking a month to accomplish the same thing?" I think you're looking at the problem through the wrong end of the telescope. The perfect is the enemy of the good.

I think by marrying mathematical concepts to computers you're continuing down a path that may no longer be useful. While it's true that computers are mathematical creations, the question isn't whether a program is "correct enough", rather whether it is "useful enough"

We're already using poor, leaky abstractions to grasp and create programs -- it's called the English language. The guys who created E-bay didn't work out the calculus of various bidding strategies -- they implemented a simple online bidding system. We're not looking for mathematical perfection. We're looking for maximum value with minimum effort.

Or maybe I'm smoking crack here. Are you saying that the ability to make the most complicated precise programs using the least amount of syntax is a good thing? If so, why would that be?

Reply on news.ycombinator.com