Without ESRAM Xb one would be at an awful disadvantage compared to PS4. With it Xb one is more comparable, but still disadvantaged.

Sure it is. MS decided for DD3 many months ago and figured out that DDR3 is too slow so they implemented ESRAM. If it helps (it sure does) than it was a decision which was tested thoroughly before with a lot of tests. The PS4 is just better with GDDR5 in such an amount but this doesn't mean MS made a bad decision at all.

If Sony didn't luck out with GDDR prices moving down and enabling parity with 8GB RAM,
PS4 would itself have had performance bottlenecks and programming restrictions with 4GB of total RAM.
(although it's plausible that the end result would at least have been on par with XBone)
The question about Sony's design would be whether they really were planning on only 4GB max,
or considered that might be their limit and thus developed for it, but actually had good reason to believe they could pull off 8GB.

The reasons MS did not go with Sony's config are legitimate, Sony just lucked out in the end, gaining a signifigant advantage.
The only other plausible configuration would be split RAM pools, which is how PCs are designed.
Both MS and Sony sought a unified memory pool for likely the same reasons: less data shuffling between CPU/GPU.
(though PS4's 50% more GPU cores with 8x GPGPU capability is vastly more able to leverage those benefits)

Excluding Sony's apparent luck with GDDR dropping in price, my only question of MS' choice would be why they went with their ESRAM solution,
instead of a PC-style split pool, e.g. 6-8GB DDR +2-3GB GDDR without ESRAM and Move Engines and instead more GPU cores,
given that the ESRAM solution forced them to remove GPU cores and GPGPU capability which would actually leverage a unified pool to the max.

Such an approach (split pools) would probably also have compared well to Sony's approach had they ended up with only 4GB total,
the benefits of a shared pool tend to be decreased if doing that gives you a much smaller total pool to work with.
So if anything, it's odd that NEITHER Sony nor MS went with split pools, even though it was plausibly equal or superior benefits.
MS clearly took a more conservative/ less 'gambling' approach than Sony, but if anything that makes me wonder MORE
why they didn't just take the truly conservative proven approach and use a split pool, for max RAM + VRAM + GPU cores/etc.
I think XBone would be looking/performing on par with PS4 if they had taken such an approach (at same cost).
So while I don't necessarily fault them for not taking Sony's gamble, better solutions at same cost do seem plausible.

Try telling that to Sony lovers. The argument will never end.

The difference between a bad decision and a good decision is time and information (incl. information about what the competiton is doing). Probably for 99% of the Xb one development DDR3 was a good decision. And it's basically impossible to change something like that with only 1% of development left. To switch to GDDR5 late in the development would have meant Xb one not launching until 2014. I wouldn't be surprised if MS sat down and weighed up delay+GDDR5 vs. 2013 launch+DDR3+ESRAM. On any given gaming forum you'd never get 100% agreement on which decision MS should have made between those, there are clearly pros and cons with both. If MS did actually think about delaying Xb one in order to put in GDDR5 but decided against it I think they made the right call. Missing holiday 2013 and giving PS4 a 6 month+ head start would have been a disaster for Xb one. As it is only launching in 13 countries before 2014 is permanently putting Xb one on the back foot in all those markets that are yet to launch. 

That never happened. Once they decided on the memory system (which likely was the second decision after the cpu decision), there was absolutely no way of changing the path. This is so fundamental in the whole process that if they changed that, they'd have to go back to blank sheets of paper.

Yes, that would be the source of a delay...

Game developers work around multiple architectures and memory systems, between consoles and PCs,

so such a delay would certainly mean a 1 year delay, but not necessarily the end of the world,

except that basic market analysis would show that Sony having a head-start would be disaster for MS,

given Sony's global demand let them come behind vs. MS initial head start (and horrible PS3 launch).

So perhaps they thought about it for a moment, but those factors would immediately rule out such a move.

I think both companies just guess at the other companies specs with a $400 budget. It didn't help that Sony put their dev boxes out with 4Gb of DDR3 memory. I would assume that MS guessed that at worst case scenario sony would bump to 8Gb DDr3. MS could have anticipated that as worst case scenario and topped it with ESram to give XBOX one the slight edge. And it would have been, and would have tipped the scales in the opposite direction. But then Sony not only jumped from 4GB to 8GB, they bought a faster ram. I think the dev boxes actually fooled MS and all developers, as they spent a year developing games for the ps4 using 4GB DDR3. All the 3rd party dev found out at the ps4 launch lol the same as the whole world. This kept Sonys big secret

PS4 dev kits had GDDR all along, just in a smaller amount.
4GB of DDR with no ESRAM doesn't make any sense.
Even embedded GPUs on PCs use ESRAM or EDRAM if they don't have GDDR.

EDIT: History Lesson: http://www.redgamingtech.com/ps4-developers-were-speechless-when-8gb-gddr5-ram-was-announced/

Ahsan Rasheed [Thuway]: “In early 2012, when the PS4 had 4GB of GDDR5, a developer in the EU region claimed: “Durango tosses the Orbis in every metric.”

Then you need to elaborate.
No point claiming someone is "wrong" and not even bothering to back up such claims, you just wasted my time with reading that for no gain or benefit to any parties involved.

No they wouldn't.
It would literally be as simple as swapping out the DDR3 with GDDR5 memory chips, Kaveri/Jaguar has dual-memory controllers which supports both memory technologies.
They might need to revise the PCB at most.

However, the benefit would potentially be minimal, you can have all the bandwidth in the world, but it means squat if the compute hardware is already fully saturated, which the Xbox One has less of.