im thinking they will spec for lockhart and then scale up for ps5 and xsx cause this is the easiest way of doing it.
Scaling down is easier than scaling up if all your target platforms share the same feature sets.
Scaling up is easier if you have disaparate features sets. For example, in the PC space there are so many different features spread across so many different hardware configurations that many developers choose to target the largest pool of hardware that share common features. Or for developers that target PC and consoles, consoles have historically been the limiting factor due to lacking advanced hardware features as a console generation goes on.
In that case it's easier to add in or tack on advanced features that are missing from what you've chosen as the base hardware feature set. Notice that I say hardware feature set and not hardware speed. You can always scale across performance targets fairly easily (within reason). It's much harder to scale across different disparate supported hardware features.
That's why DirectX allowed PC gaming to explode like it did back in the late 90's early 2000's. Programmers could code to a common hardware feature set and then scale according to speed. It wasn't perfect however, as each hardware still had their quirks and then you have the infernal hardware feature support cap bits for non-base hardware features.
Contrast that with say historical cross platform development on console where PS2 was nothing like Xbox or Dreamcast. Same goes for PS3 being absolutely nothing like X360. PS4 and XBO got pretty close but there were still some big sticking points of hardware divergence (ESRAM?).
In the case of XBSX and PS5, with relatively minor differences we have very similar hardware feature sets. With XBSX and XBSS we have virtually identical hardware feature sets. That's about as easy as easy gets in terms of scaling across different performance targets.
What does that mean? It's easier for a developer to scale down than it is to scale up. In other words, it's easier to create rendering and assets targeting higher fidelity and then decrease that quality for a lower performance target. Scaling up would require starting with low quality assets or rendering and then somehow figuring out a way to add in higher detail to base low quality assets. *buzzer sound* Nope, easier to scale down.
And this is made possible because you don't have to worry about whether the lower target supports X hardware feature or can do Y rendering method.
Regards,
SB