Of course the Archicad is not dying.
It has ingenious features and software philosophy.
But GS had some bad decisions in the past, about the development of the software, I think (because of the urge of the yearly new vesions).
Instead of turning archicad a open platform of designing architecture, they introduced some partly perfect new tools (starting from the curtain wall, the shell and the morph tool). New tools, instead of simplifying and reducing the number of the existing tools.
E.g.: I don't know why there are separate tools for windows and doors (both are openings, with almost the same working manner).
And more about the openings: why should I place the doors/windows only in walls? They are same gdl scripts, like the objects... Why there is a different tool for lamps? (with only a few more parameters, than the normal objects; and with the user interface scripts you can do any kind of parameter adjusting panes)
Why there are differencies between the slabs, roofs, and even the 2d fill tool (all of them are closed polygons, with some additional parameters) etc.
And about the open platform of architectural design:
GDL is good. It was revolutionary, when it was introduced. It was the main strength of archicad. The ease of scripting a new object. But it was decades ago. The GDL represents the coding paradigms of the times 40 years ago! If you want to change a single parameter, you have to enter even 50, consecutive, comma separated values; instead of saying what parameters you want to adjust. And the gdl editor doesn't have any syntax highlighting, or any syntax aids.
In the past 40 year many new paradigmatic changes occurred in coding. E.g. the object oriented programming.
I think the GDL needs a complete revamp, according to the followings:
1) the new gdl (NGDL) should control (almost) everything (not being just a closed container);
2) the NGDL should be completely internet based (and web compatible - like: svg, and JavaScript);
3) the NGDL should be object oriented;
4) the backward compatibility should be handled in the archicad, not in the NGDL script language;
5) the NGDL should have a grasshopper-like graphical interface;
6) the NGDL should be an open source platform (with ease of sharing new features, tools, and scripts).
A feature like this could solve some old-old problems, like stair-maker's, and the lack of some features about the existing tools.
I think the main miscomprehension in the GS HQ is that this would ruin their business. And their 3rdparty partners' (like Cigraph, cadimage etc.) businesses, too.
No, it wouldn't ruin them, at all. But contrariwise: it would give a huge strength.
Strength of the community of self driven, volunteering (and for profit, 3rd party) developers. Strength of an universal, open software.
And it wouldn't mean, that they should stop developing Archicad. Absolutely not.
Archicad needs a lot of developments, smaller and major ones, too.
E.g.:
-UI developments, based on UX paradigms (even some special hardware input device development would be great - I don't think that the mouse and keyboard is the ultimate input devices for architects);
-performance improvements (like the speed improvements in the last couple of versions);
-BIM, BIMX, IFC, OpenBIM, and other interoperability improvements;
-minor bug fixes (there are several annoying small bugs - which are really annoying, but small enough, not to report them), it's a hard work to detect and collect these small bugs.
All of these would give enough work to do for the forthcoming 10-20 years; and GS could come up with new features, in every year. The archicad would definitely remain the quasi standard platform of digital architecture; with its outstanding core features of editing, virtual tracing, publishing, teamworking and the open source digital architecture community.
Talmácsi, István, architect (AC user since 1997, ac4.5 - now: ac18)
