You suppose incorrectly.
But if we are making ungrounded extrapolations from a single observation, let's go the other way:
I may prefer to play a rogue (assassin) 4/rogue (thief) 4 than a rouge (either) 8. How is that fundamentally different from the possibilities that you suggest, except for an arbitrary line-in-the-sand? Or a Cleric (War)/Cleric (Light)? Eventually, a line has to be drawn as what is a class, what a subclass.
Illusionists and Assassins have been independent classes in the past, now they are not. We roll with it. But the choices made by designers should consider not what one individual somewhere might someday want to play, but what makes the game that meets their criteria of success (including crucial but unquantifiable virtues such as marketability, fun, "balance", etc.).
I don't think it's possible to argue on the "fun" aspects here, because it's hard to tell what the majority prefers between one arcane class to rule them all VS multiple arcane class. I want to point out, that we still have multiple divine classes, multiple martial classes, multiple hybrid martial/divine classes... in your previous post you seemed to like having one arcane class only, but now you say I am wrong assuming you want the same for other classes groups. Still, I won't believe this is the will of the majority. I would rather bet that most people prefer either few wide classes or many narrow classes, but not a system where one group (arcane casters) gets a one wide class implementation and other groups (divine casters, martials, roguish types) get a different design approach. Anyway, at the same time I don't think WotC is simply trying to get the will of the majority, because they probably realize there often isn't a majority but on most topics there are only minorities, and they want to satisfy as many of them as possible.
Instead, let's argue on how the mechanics allow you to play your wanted character. Still somehow talking about fun, balance and marketability, but trying to focus on how the mechanics
enable those.
You want to play a character that is designed around 2 conceptual elements: (1) assassin, (2) thief. Can you do it with the current rules? Not well, because both are well represented by the Rogue class but are mutually exclusive choices (although another player might be satisfied by picking either subclass and complement it with feats, let's assume here that you don't find this satisfying enough). So in this case, the current mechanics don't support your desire. BUT there is a way out of this, because Mearls announced that the Advanced game contains official rules for creating your own subclasses, and that includes mix-n-match existing subclasses, which is actually easier than making one from scratch. In this specific case, you'll probably pick some subclass features which best define you as assassin, and pick others for thief, result may not be perfect but can look VERY close to what you would get if you allowed multiclassing of the two. Just looking at the fact that rogue subclasses are a bunch of 6-7 individual features, it will probably end up being a simple matter of picking as many of them a-la-carte. With this in mind, you can be very optimistic that even if there isn't a hybrid assassin/thief subclass, the game supports you in making your own.
Let's say there is someone else who wants to mix 2 arcane spellcasting classes. Here the job might be much harder, because what makes them different is not just a bunch of features, but entirely different spellcasting mechanics. It's not you can easily mix them with an a-la-carte process here. Instead, we have a framework for mixing 2 spellcasting classes, provided in the form of multiclassing rules.
I imagine you can say, that you can then solve this problem by applying multiclassing rules to subclasses of the wizard class... then there is not much difference between having sorcerers and warlocks as subclasses or classes, except for the
presentation. That includes for instance the question, do we really want to
present one only arcane class with 100 subclasses released in the first e.g. 3 years of splatbooks, while we have 5 martial classes with only 20 subclasses each in the same time span? Wouldn't it be better to split the arcane class, like it was in the last 2 editions?
There is another underlying problem, the fact that different gamers want different spellcasting
flavor ("I learn magic from books" vs "I have magic blood/soul/other bodily fluid" vs "I get power from a pact") and others want different
mechanics (Vancian, hybrid like the current rules, at will, encounter-based, spell points...), but then they may also want different
combinations of their 2 favourites. Someone wants a book-wizard but hates vancian, another wants a dragonblood sorcerer but with spells points...
WotC is here in the difficult position that, if they link a spellcasting mechanic with a flavor concept then some players will revolt that they can't play what they want, if they don't link spellcasting mechanics with flavor concepts then some other players will complain that there is no fantasy world consistency. I don't know what would be best, but I bet they are going with officially linking mechanic with flavor, and then just suggest for individual gaming groups to change the flavor on their own. Let's keep in mind that flavor also ties the classes to the fantasy world setting, so it has implications on supplements and published settings as well (e.g. if you have a class, people will expect clear support for that class in supplements, if a class is turned into a subclass then the expected support is less) so it also depends what are their plans for future developments on the Sorcerer and Warlock concepts in 5e. If they have few ideas only, then they may just go with subclasses.
With all these in mind, my preference would be for separate classes (especially because of the presentation issues, i.e. having multiple arcane classes like divine classes and other groups, having reasonably equal treatment/support for each class) but clearly the issue is very complex.