Just to remind everyone of old news, seamless branching is not something that can be easily done at home. In fact there doubt as to whether it can be done at all. Yes it's not so complicated that you can't back-up DVD-9 seamless branching discs onto DVD-R, but from what I've heard it's so complicated and difficult to do that it is simply not worth the effort. The 3.5" floppy disks today has not changed in more than a decade. Despite this Microsoft's format has not changed. Yes you can format it in Fat-16 as apposed to Fat-12 which was possible some years past, however Microsoft simply does not do the best it can with its format. There were various other formatting tools available for DOS which would actually allow you to format your disc so that the end of each track aligned better with the start of the next one, meaning that when it spun and was read, the read-rate was the same but the idle-time was significantly shortened, which then in-turn created a disc that would appear to read more than 50% faster than a standard floppy.
I believe some people confused this old concept with the idea the same thing would have to apply to branching (that at the end of the "track" should be logically aligned to the next possible branching places) - however I don't think that is the case. The problem doesn't seem to be accessing the next information fast enough, the problem seems to be in the MPEG format which was never originally designed for branching. Yes it can be done, but no the DVD format wasn't specifically designed to allow for it. The floppy disk wasn't specifically designed to simply "slide" to the next track efficiently, yet under the right formatting parameters it is possible (just not with Microsoft's format). And it may not make much difference today, but you think about all the software that used to come on half a dozen, or more, floppy disks - that they ALL could have loaded more than 50% faster if IBM had designed the floppy disk to take advantage of known hardware capabilities, and the position of the head when it reaches the end of a track. See the problem was the standard thing to do was to start the next track the same way as the previous one:
1_2_3_4_5_6_7_8...18 (first track)
1_2_3_4_5_6_7_8...18 (second track)
Now the first track ends pretty much where it started, the head then moves down to the second track. But hold on, the second track already started. Because the disk is already spinning it's missed some sectors, let's say it misses the first 3 sectors and starts at 4 - that means it has to make a complete rotation back to 1 to begin reading the second track. Alternatively it could have been formatted to take full advantage of the hardware by starting with the last 3 sectors... and number 1 where number 4 would have been. That way when the head moves across to the next sector it's on number 1 right away and doesn't need to waste time rotating.
Such:
1__2__3__4_5_6_7_8...18
16_17_18_1_2_3_4_5...15
Yes this does not cover the full scope of how a disk is physically made up, but in simple terms that's how it works. Like I said the format ALLOWS for it, but it wasn't DESIGNED to cater for it.