What you are doing isn't a single step PowerPoint to PDF and Ghostscript is not rendering the PowerPoint. In fact if you are creating a PDF file Ghostscript isn't (ideally) rendering anything.
What's actually happening is that you are asking PowerPoint to print to a canvas, which is then passed to the PostScript printer driver. That produces PostScript which is sent to the Port. Your (and others) Port Monitor then sends the PostScript to the 'Distiller' (in your case Ghostscript and the pdfwrite device). The Distiller reformats the vector drawing commands into a PDF format and builds a PDF file from them. It doesn't render (turn into a bitmap image) anything unless forced to.
Obviously there are several places along that road where the problem could creep in. Given that you say that the Adobe product (the others you mention al use Ghostscript) has the same problem, I think its safe to assume that the problem isn't Ghostscript.
This also means that you aren't using the driver you think you are. Adobe can't handle PCL as an input medium as far as I'm aware, and nor can Ghostscript. GhostPCL will handle PCL as an input, but that's not what you say you are using.
Of course you haven't linked to an example file to demonstrate the problem, nor supplied an example command line, so this is all supposition.
Now if, somehow, you are using a PCL6 device, then the problem is most likely due to the presence of rasterOps in the output. Rasterops are part of the PCL imaging model which do not exist in PDF and are a form of transparency. There are three ways to handle such content for a PDF output device; firstly render the whole page content to an image, secondly ignore the rasterOps objects, thirdly treat the rasterOps as opaque.
GhostPCL and the pdfwrite device take the third option. So, its just conceivable that your original content has some transparent objects which are being handled as rasterOps by the PCL printer driver, and then rendered as opaque by GhostPCL and the pdfwrite device.
If that's somehow the case then the solution is simple; don't use a PCL printer driver, use the PostScript one.
If you post a link to a (simple, eg single page) example of what you are sending to Ghostscript, and a command line, then I can look at it. Please don't send me the PowerPoint, I can't use it and even if I could, my print setup would not match yours. I need the data being sent to Ghostscript.
[EDIT after looking at files]
Don't mean to sound like I'm lecturing, the problem is people find these result on Google searches and then try to apply them based on a poor understanding of what's happening. So I find it best to be really clear in my answers about what's going on. It saves questions later :-)
The first thing I see is that the PCL is indeed PCL, and if you try running that through Ghostscript it throws horrible errors and exits. So presumably you aren't doing that.
The PostScript file contains nothing except huge images, rendered (presumably at 600 dpi) contains 2 pages, the two pages look like your images above. Which is why the PostScript is better than 20 times larger than the PCL file.
But.... If I open the .ppt file with OpenOffice (4.0.0 is what I have to hand) I see exactly the same thing. I don't, I'm afraid, have a copy of Microsoft PowerPoint, but from what I see here there are two conclusions;
firstly that the PDF I get looks pretty much like the PowerPoint when viewed with OpenOffice at least. So there's something 'interesting' about your PowerPoint.
secondly, even if that's not what you expect, its what's in the PostScript program. That means that either PowerPoint rendered the slide to a bitmap or the Windows printing system/HP driver did.
Now, if I run the PCL through GhostPCL instead of Ghostscript (rendering, not producing a PDF) then the result is more like what I think you are expecting. However, when sent to a PDF file the result is horrible. Which strongly suggests to me that there's some form of transparency involved, PostScript doesn't support transparency at all, and PCL does it through rasterOPs.
I'm afraid that this means that the problem lies either in PowerPoint, the Windows print system or the PostScript printer driver you are using. Since the PCL is at least close to what you expect, I suspect that this means PowerPoint is doing the right thing, and its the printer driver messing up. It appears you are using the Windows PostScript printer driver.
So there's no way you can 'fix' this for files like this, at least not with Ghostscript. You would need to 'fix' the Windows PostScript printer driver, or possibly the Windows print system. You could try reporting a bug to Microsoft, presumably these files print incorrectly when sent to physical PostScript printers too.
Output From Ghostscript in PDF Below
