I do NOT want them to "spread out" first level abilities or delay them for later levels to inhibit "dipping" the way 3.5 did. I never saw the problem with dips. It's not like it hurt anybody's feelings that somebody took 1 level in another class.
Among the top 3 reasons that makes me hate 3rd edition with such a fiery passion is that they made you play a character for weeks or months before he could actually do what you wanted him to do.
No more.
Well, I don't like that a player can just dip and pick up a new class. The problem you mention from my perspective was bad class design and a focus on Prcs rather than class variants.
a. The designers building to many prescribed abilities into a class (a mistake which they are, in my opinion, repeating). When the designers force too many specific class abilities, you alienate many players, because you are forcing a single vision on the player and DM and abilities that may not fit. This is a big issue that I have Next's classes;
b. Didn't have enough common fantasy archetypes covered in the core book including: Arcane Warrior, Barbarian (Wilderness Warrior rather than Rager), Berserker (urban), Bard (Divine), Bard (nature), Cleric (non-martial), Monk (Arcane), (Divine), Psychic, Psychic Warrior, Ranger (urban), Ranger (non-spellcasting), Rogue (martial), Rogue (wilderness), Shaman, Swashbuckler, Witch
c. Until Unearthed Arcana, focused on PrCs rather than class variants at 1st level. Class variants were discussed in the 3.0 PHB and the 3.0 DMG (variant spell lists, variant skills, etc.). However, lack of examples and support led many DMs to ignore them.
The result was multiclass hoop jumping that I hated as a player and DM. As a 3e DM, I used several third party classes and class variants to help my players avoid the multiclass hoop jumping to create common archetypes (provided they fit my campaign) so they could play them at first level and reflect their character's initial training.
Additional class, better class design (which would involve less prescribed abilities and more choices), and decent paths/orders/schools, etc. would, in my opinion, be a better solution to multiclassing for common archetypes that are viable at first level. I would save multiclassing for picking up a new class after first level and that, in my opinion, that should require prerequisites and require picking up individual class features more slowly/organic.