It's not that the file format was any better, it was just much simpler because in the DOS days you couldn't arbitrarily embed content from one application into another. Lotus' file format was just as proprietary as Microsoft's was. However, Excel had very good file importers, but also things like keystroke migration wizards etc. Microsoft invested a lot of time and money into doing everything they could to lower the transition barriers into Excel specifically for Lotus users. (and Office apps in general. They had similar efforts for WordPerfect->MS Word migration and others)
It was a fairly one-way trip in that even in the first versions of Excel, getting data back out was never as easy as getting it in was (i.e. the exporters weren't as complete as the importers... probably by design) and throughout the 90's Microsoft definitely built on and expanded the one-way nature of migration to their proprietary file formats as a key to lock-in. It wasn't so much that it was all that difficult as it was undocumented and always changing (there were multiple backwards incompatible file formats in each of the main MS Office apps in the 1990's). So if you were a customer or competitor, it wasn't so much that you couldn't reverse engineer the file format, it's that they were relatively undocumented and constantly changing so by the time you had done it, they were on to the next revision. I think it was a couple of years after the DoJ case, and a result of the settlement terms, that things really started improving on this front... but by then Microsoft had long since won. By that time most of their competitors had already figured out that the only way they could hope to compete with Microsoft was to bypass them entirely... hence the focus on web apps.
You mention web apps so I'll close with this: unless you're taking great pains to only use the open and cross-platform standards for them, in the not too distant future you're likely to find yourself in the exact same spot with your web apps that people were with desktop apps not too long ago. Different platform, same game...
I wrote a bit about this in another place, but the key to understanding the situation is to understand that the file format does not actually tell you how to format the text. Especially in the early days when rendering a font would work completely differently in each application because you never knew what font metrics they were using.
It's easier to be second to market because you can copy the leader. In this case, if you write your formatter with the intention of importing documents from the leader, then you always have a target. If you then add features which the leader ignores, suddenly they have a much more difficult time to catch up -- their formatter is not purposely built to render your file format.
As I said in my other message, I've completely forgotten the Excel file format, but Word's format was originally just a binary dump of all their data structures. That's why it is insane. It's hard to read unless your code happens to be structured that way.
Anyway, I'm +1000 with you on the importance of open and cross-platform standards. Just think about all those stupid rails apps -- your data model looks exactly like your sql tables. Then you dump those same stupid tables into json and splat them over the wire for someone to try to make sense of. Deja vu all over again ;-) I suppose I should be happy it's not XML...
So I guess we both agree that the resulting complexity of the file format itself had a lot to do with an intention of lock-in. Which directly contradicts Joel's first statement: "are deliberately obfuscated". :-)
Also, regarding web apps, I like to think of it as the building, not of "perfect systems", but "self-correcting systems". The web, with all its chaos, it still more self-correcting than MS Office has been - in fact, even from your comment it does seem that the intention was only lock-in and not improvement of file formats (other than perhaps an improvement in MS bottom line).
Oddly, even the benefits to the MS bottom line now seems more short sighted. E.g. the ongoing fiasco that is the Windows 10 auto-upgrade. [1] Someday I expect to read post-mortems on the failure of Windows 10's so called perpetual upgrade policy (i.e the notion that there will be no more version numbers after 10) which link the failure to their refusal to build self-correcting systems.
[1] Personally, I don't think we have seen the worst of it just yet.
I'm posting too much on this topic :-) Just a quick word from someone whose job it was to write filters for these file formats. They are not deliberately obfuscated. They are just the product of long lived legacy projects with short deadlines and poor process. It is absolutely obvious when you work with the formats that this is the case. Word in particular was clearly just a dump of the internal data structures in the app. It's not even that complicated -- the complicated bits come later and they are all for either backwards compatibility or for the fast save feature (or whatever it was called). The fast save, in particular, is complicated and is that way because it is append only. It is essentially a series of patches on the original data structures. Yes, it will make you want to rip your hair out, but it is very, very obvious why it's that way.
It was a fairly one-way trip in that even in the first versions of Excel, getting data back out was never as easy as getting it in was (i.e. the exporters weren't as complete as the importers... probably by design) and throughout the 90's Microsoft definitely built on and expanded the one-way nature of migration to their proprietary file formats as a key to lock-in. It wasn't so much that it was all that difficult as it was undocumented and always changing (there were multiple backwards incompatible file formats in each of the main MS Office apps in the 1990's). So if you were a customer or competitor, it wasn't so much that you couldn't reverse engineer the file format, it's that they were relatively undocumented and constantly changing so by the time you had done it, they were on to the next revision. I think it was a couple of years after the DoJ case, and a result of the settlement terms, that things really started improving on this front... but by then Microsoft had long since won. By that time most of their competitors had already figured out that the only way they could hope to compete with Microsoft was to bypass them entirely... hence the focus on web apps.
You mention web apps so I'll close with this: unless you're taking great pains to only use the open and cross-platform standards for them, in the not too distant future you're likely to find yourself in the exact same spot with your web apps that people were with desktop apps not too long ago. Different platform, same game...