Historically, XFCE had more manpower and resources.
Started as a small project, growing slowly, and finally became a major DE.
XFCE even underwent some complete rewrites.
The first version of XFCE in 1996 is not written with GTK+ but in XForms library.
It's written with their own toolkit, and later they ported it to GTK+.
After 16 years of developement, now it has became very mature.
When we started LXDE at the end of 2006, we only had a file manager + panel.
XFCE was already in a quite good shape, having some developers, and
included by major distros. Not until 2008 did LXDE became relevant.
So the comparison, to be honest, is not that fair.
But this is not an excuse and indeed we should do better.

The problem of the project is clear. It's too centralized.
In the old days, I maintained more than 70% of the project.
So after I graduated from the university and got a really exhausting job, many things became nearly unmaintainable immediately.
Luckily, we got a new member Marty Jack.
He did quite a lot of things to keep the project alive, especially the desktop panel.
Due to the divergence of design choices and lack of help from others, he left the project later.

Fortunately, the situation is better now.
The file manager and its core are now actively developed by Andriy.
The session manager lxsession is now developed by Julien.
The translation part is handled well by brother.
For other parts, we tried to reuse existing technologies if they are not too heavy.

Most parts of the projects are already compilable with both gtk2 and gtk3 now.
I tried to update lxtask and developed a replacement called lxsysmon based on gtk3.
However, after testing, I found it more resource-hungry and some problems were caused by performance degradation in gtk3. So I stopped that work later.

The most problematic one must be the desktop panel.
Previously I worked on it and started lxpanel2, a rewrite in gtk3.
I, however, got stucked a little bit when working on multiple monitor + multiple panels support. Because it is not working correctly, the code is not yet pushed to git.

Currently my plans for it are:
1. Finish the multi-monitor support. I tested several other panels and made a new design decision. The original plan for panel positioning/sizing model is too complicated and hard to implement. Let's make it as simple as possible. I'll find some time to do it recently.
2. Support for pseudo-transparency is removed. I plan to use X composite instead. (of course a composite manager should be running) Pseudo-transparency is inefficient and may causes memory leaks sometimes.
3. Support basic theming to some degree. GTK+3 has much better css-based theming support than old gtk2.

Regarding to the image viewer, I'd suggest replacing it with some other similar programs as it's not maintained now.

The calculator "galculator" is updated and now supports gtk3.

For the text editor leafpad, we may try "l3apad", a gtk3 port.
http://gtk-apps.org/content/show.php/L3afpad?content=144892

The window manager openbox still works well and since it's not maintained by us, we don't need manpower for it.

These are the current status of the project. About what's lacking and possible solutions, I write them in another mail.