Essentially time and cost? Building such a Frankenstein-Phone would probably take you half a year to design the PCBs, get all the connections and power rails right, all the needed peripheral electronics for the chips. Read thousands of pages of datasheets to place the correct capacitors for the oscillator of the ... sensor on your mainboard design. (And there are a lot of tiny components in a phone that all work together, in part depend on each other, or require additional control/supply circuits.) You'd need a lab and equipment do build it, and the mechanics and encasing. And probably some takes and failed iterations. And software and drivers also have to be rewritten and patched.

So I'd say if you have the expertise in electrical engineering, hardware design, embedded software programming... A 5 figure(?) sum of money for supplies and equipment and nothing to do in the next year... I'd say nothing is stopping you 😆

There isn't, like, a standard like there is for PCs. Everything is custom made for that thing except the microchips themselves. And taking those off and just slapping them on a different board isn't that simple.

I would love it if they were like a PC though. Build my own phone with this screen and that CPU/GPU, configurable RAM, etc. Why they don't do that, though? Greed.

I'd say that I could do it for you, even, given sufficient effort, time and money. However, it would be the size of a shoebox. And don't you dare open that shoebox or else all the parts are going to come falling out.

The most modular phone right now, which you can open yourself and replace parts with just a screw driver, is the FairPhone. And even that one, you can scavenge parts from older models of the same brand, because the connectors don't fit. There's very little space left inside a modern phone.

It's prohibitively difficult. Like not just hard, but complete redesigns for even fairly small adaptations.

Hardware: The parts just don't replace each other one for one, and it's not just where the wires are. Each SOC or component requires a cadre of resistors capacitors, voltages, signal lines that don't line up well between different products. The boards that these components mount on are many layers thick with wiring hidden in multiple layers. You can't even just bodge and reroute everything all the time. In many cases the packages wouldn't even fit in the intended target spot and phones have precious little space to spare. Then for a lot of chips you've got thermal considerations.

The 10x camera for an s23 wouldn't have a chance at fitting in a pixel, the focal lengths are different It literally wouldn't even fit in the case.

Drivers/software: especially relation to cameras, a lot of third party software can't even run the full compliment of Samsung cameras. Commands to switch back and forth between lenses aren't universal. In a lot of cases you can screw around with different camera software and get it to work and make modifications for the cameras to signal when they need changes. But then when you take a s24 which is 64-bit only you can no longer run any of the 32-bit camera software that used to do things like sphere camera. And then even if you did manage to swap sensors out all of the lens correction would be wrong. You'd end up with Chroma and correction issues. The cameras aren't just giving you what's off the sensor anymore. A long time ago we used to get big upgrades in picture quality going from one sensor to a new more sensitive sensor, We now do the opposite and use really big sensors that take multiple samples per pixel and we drive those pictures with complicated software or even AI to generate better looking imagery. When all the software and hardwares tuned together to give you a better image and you swap the hardware out it's a bad thing.

These devices are all very custom they're very purpose-built around each one of the features and the subcomponents don't even match up neatly in between different models in the same line. I'm fairly certain you couldn't even fit an s23u 10x camera into an s24u.

TL;DR it's essentially as close to impossible as you could make it in just about every way you can imagine.

Imagine you like the shape of the front of a Mini Cooper and the rear of a Ford mustang. You could take the paneling from the interior of a rolls Royce and the seats from a Lamborghini and make a really cool car.

Unfortunately, unlike modern standard PCs, phones are individually designed and built and even models in the same range can’t use each other’s parts or software.

Each component is designed to work with each of the other components and just slapping them together doesn’t necessarily make a new working product.

design compatibility issues and proprietary firmware or software

All of this stuff is usually fit onto a single board, crammed into a very specific amount of space, and is thoroughly and iterated until it works properly. This isn't the kind of stuff a home lab does, but you could certainly try. I think it would be damn near impossible to do it better and more reliably than teams of hundreds or thousands of various engineers. It's not like you can just take a phone CPU and slap it on a random board without a ton of forethought.

There were modular phone projects that were killed by google.

But it's intentionally hard to do it otherwise, to make more money out of broken phones.

Software.

You won't have any of the electrical or protocol/register info or other data for any of the components unless you're a manufacturer and most parts aren't really salvageable separately but are essentially one big glob on the board. Even with the skills, you'd need to reverse engineer some of the most complex and hard to use components ever manufactured for consumer use and somehow fit them in places they were never meant to fit.

And yes, software. The board support for the SOC, mostly. Maybe starting off with a pinephone or something might help, but I doubt even that is open and usable enough.

One time I took apart my Nintendo DS. I broke my Nintendo DS.

Another time I took apart my PS2 controller. I broke my PS2 controller.

Now you want me to scrap parts from a phone, and build from scratch??? Aw hell naw!!!

As long as you're not selling the phone afterwards I'd say the only thing stopping you would be the skill and knowledge required to do such a thing and the necessary tools.

Keep in mind most of the parts you're talking about are to some extent proprietary and specifically designed not to be cross compatible, so that may be a big hurdle

Modern phone cameras are 99% software at this point. If you took one phones camera and put it onto another phone without adapting the software then the photos will look like ass.

You can’t just unplug something from device X and connect it to device Y. The connectors aren’t standardized. And even if they were they wouldn’t fit because the placement is different. In theory you could take the CPU off of one device and plant it onto another. But have fun with that BGA micro soldering. Plus the connections will be different unless you picked a phone with the exact same CPU. Ram is the only thing you could potentially upgrade. But like good luck.

You’d legitimately have an easier time making a new phone from scratch than trying to piece together 3 different phones.

why aren’t we able to just make it more to our liking?

You are, but you have to figure out how to get it all to work together yourself. That means physically, electrically, and in software. All the things you listed are problems that need to be overcome. If you have the ability, you can do this, but you yourself have to do it or find someone who specializes in such particular work. Most people have neither the skills, time, or patience to do this.

Don't forget about device drivers. I can't even install a newer version of Android on my Android phone because the community never managed to get the antenna to work after upgrading the OS.

I can't solder nearly as well as a machine.

Form factor, mostly.

Capitalism, kind of. Practicality, the rest of the way. Mobile parts need to be designed really tightly integrated, because they need to fit exactly into such small spaces, and standardizing them isn’t really feasible without significant pressure on the market (aka, socialism).

The reason desktop PCs can be so standardized is how big they are. Tons of room for customized parts.