Why does it need its own site?
I believe this topic is broad enough and specific enough it deserves a separate site.
Questions about specifics of "deeper" embedded programming usually fall on deaf ears when asked on StackOverflow, or are met with answers from users clueless about it (suggesting a JBOSS solution for a tiny embedded ARM9 platform?)
There are subjects that go entirely cross-site: Take this example: "How to perform the fastest bitwise XOR in 8051?" There are some smart assembly language answers but the best answer is to attach a 74L86 or similar to an IO port. Who on StackOverflow would give such an answer? Or what about creating a driver for a device? How much to do in software and what should be pushed to silicon?
There are subjects that would be really hard to fit anywhere: how to implement a trivial filesystem matching specific needs?
Matters of getting crosscompilers to work with a given chip are a terrible fit for SO, poor fit for Serverfault, very poor for Electronics and completely wrong for Programmers.
Questions about using CPU software features like FPU on a given ARM chip are a very poor fit for Electronics and too specific for SO.
In essence, just like Security.SE is a mix of Superuser, Serverfault and SO, or as Gamedev is just a subset of SE, Programmers, Careers and Writers, Embedded straddles SO, Electronics, Programmers and a good deal of content too specific for any of these sites.
As for fitting Electronics, I suggest some searches:
Essentially, the "High Embedded" - Industrial, Linux, RTOS, while allowed on Electronics, is an unwanted child. It's too far away from iron to be of interest to bulk of users, and the bulk of diode-MOSFET-supercap questions scares any professionals willing to discuss them away. Even programming close to iron is a margin. [Assembly] gets 49 hits.
And in the end, let me dedicate this closed question to anyone with doubts, with its comment:
I suspect that an Embedded Programming and Design site would have very similar problems to the ones that Robotics is currently having.
It too is a site which brings together elements of Stack Overflow, Electrical Engineering, adding in elements of Mathematics, GIS, Computational Science # and mechanical engineering as well as things which are highly specific to Robotics itself. # In beta for 425days and counting.
The problem is that we have a real problem getting enough good questions when people have so many other places to ask. Finding enough Embedded Systems questions which aren't already explicitly on-topic for Stack Overflow or Electrical Engineering and can't be answered by those much bigger and more established communities will be very difficult. That's no reason not to try, but is something everyone needs to bear in mind.
Looking through the current proposed questions, I see none which shouldn't be welcomed on Electrical Engineering, (given the embedded, microcontroller and various architecture tags such as arduino, pic and arm) but if a culture is starting to develop to close such questions then it might not always be the ideal place for them.
On another point, you have to be careful not to have a build it and they will come attitude (which is common here on Area 51) , since it is not enough for a successful Stack exchange site, you need to start planning for promoting the site within the target community now, and getting a head start on answering the The 7 Essential Meta Questions of Every Beta right at the definition stage of an Area 51 proposal.
These essential questions are:
Obviously 2, 3, 4 and 6 can wait, but the sooner 1, 5 and 7 can be answered, the stronger this proposal will be.
Ultimately, if you want veteran embedded systems controllers to move from Stack Overflow and Electrical Engineering to a vertical site then you are going to need to provide something new, different and unique, and this needs to be demonstrated by providing proposed questions which are not already on-topic on those other Stack Exchange sites.
This is something I struggled with, and I came to the conclusion that while there are some aspects of embedded systems that are covered by Stack Overflow, and some that are covered by Electrical Engineering, there are also aspects that are covered by neither site.
EE.SE is centered around electronics and has a difficult relationship with software. While the FAQ cites “the writing of firmware for bare-metal or RTOS applications” as on-topic, in practice, there are very few questions about software on EE.SE. So I asked on Meta.EE.SE, and the dominant response is that
There is a gap between SO and EE.SE when it comes to the kind of low-level programming that firmware and driver writers do. Even subjects that are technically on-topic on both sites tend to be poorly received.
EE.SE cannot engage people who work on the software side of embedded system, because of their focus on hardware; nor does the EE.SE community want to reach out to programmers and broaden their site.
It's also difficult for an embedded developer to feel at home on SO, because there is no good way to filter interesting questions. The way to filter questions is with tags, but the embedded tag is a hodgepodge of questions that may be about embedded systems or about embedding some piece of software in another. The moderation on SO is rather off too: like all esoteric topics, a question that your average C or C# programmer doesn't understand is likely to be closed; unlike esoteric languages, embedded questions do attract attention outside the embedded community.
In addition, there are subjects that are squarely off-topic on both SO and EE.SE, but part of the job of an embedded developer. Subjects like how to boot a chip, how to connect a debugger, how to interact with a ROM are off-topic everywhere.
There are precedents for sites that have some overlap with SO but cater to a domain-specific audience and cover more than programming: Game Development, Database Administrators, Programmers. Embedded.SE would be another site of this kind.
There was a prior proposal for a site about embedded systems. It was closed and deleted because it couldn't reach enough committers after two years. Given that Embedded.SE would not be a duplicate of any existing site, the only reason not to create the site would be if it again fails to attract enough participants. So the proposal must remain open.