Perhaps if analogies don’t help, let’s try pictures! Suppose the denizens of Plato’s cave have only ever seen shadows on the wall. Of these, they can take two measurements: height and width.
In the image, Arrow A1 has a height of 6, and a width of 1, while A2 has a height of 8 and a width of 7. A2 therefore has a greater height than A1. Furthermore, the cave-dwellers hold the following to be true:
Of two objects of unequal height, one (and only one) is always higher than the other.
But now suppose, perhaps due to some looseness of the restraints, one caveperson finds out that they can tilt their head to some degree. To their astonishment, they will find that now, all of a sudden, the heights and widths of objects—thought heretofore constants by all of cavepeople science—begin to change! Moreover, if they tilt their head far enough, the relative heights of objects may change, too—what was of greater height before, now is of lesser!
Now, arrow A1 has a height of (about) 4.3, while A2 has a height of 2.5. Thus, now, A1 is higher than A2!
Then, suppose they try to tell their spelunked compatriots of their findings. One pipes up, and says—“If what you say is true, then this would mean that arrow A1 is shorter than A2, and A2 is shorter than A1. But this is not possible.”
This is the argument you’re making. This is also why everybody is telling you that you’re assuming a third frame of reference: it is the case that in one particular frame (that is, at one particular tilt of the head), arrow A1 can’t be both shorter and longer than A2. But the claim that is being made is that there is one frame (i. e. head-tilt) where A1 is shorter than A2, and another where things are the other way around—which is trivially possible.
The problem is that we, and the cave-people, are used to thinking of two parts of a unified quantity—width and height as parts of total length in the caver’s example, and time and space as parts of the space-time interval—as separate quantities; but they’re not, but rather, they’re intermixed by a particular transformation (a rotation for the cavites, a Lorentz transformation for us). Total length, for the caveys, is invariant; space-time interval for us. But neither time nor distance / length are, on their own.