An Agent can do many thing on behalf of a character, and we all know of the "hacker-in-a-box" solutions that can be made.

But what stops you from firing up your Agent rating 6 and asking it to write a new program for you?

I mean, as long as the commlink or Node its running on is up, wouldn't it just patiently code away until completed?

How does a character do it? - They know the intended effect, but other than that, they are very likely working within the capabilities of the programming tool and following rather established methods.

So I suppose my question is, do the metahuman component add something vital or can expert systems maintian themselves at SOTA and even generate new programs if given enough time?

And as soon as you say yes to this premise, what do developers do then? - Supervise their mechanical codeslaves?

The basic reason I have, which serves to explain a lot of stuff in the SR world of hacking, is that hacking code is a lot like modern day viruses. If it gets spread all over in a short period of time companies will have clamped down on it.

That's why software cost so freaking much instead of there just being free rating 6 programs for all. As soon as you give everyone a copy of your stealth program companies will offer a patch to analyze and you are shut down. It's also why hackers are all about writing their own code, and why there are so many software companies.

I'd consider that agents would tend to make code that would already fall under many existing patches and generally underperform.

Cuz that's a task for the Software skill, and agents don't have that skill (p. 213, SR4).

We can do it now by genetic programming, so yes.... but it'll take a long time and be very inefficient initially.