Where will the social developers code? And what?

Dion Hinchcliffe wrote a very interesting piece, and I missed it. But thanks to John Rymer I picked it up.
John shared an interesting question indeed:

Client gives up on enterprise PaaS deal with #Salesforce.com, asks me for options. Focus is "systems of engagement." Interesting question.
— John Rymer (@johnrrymer) February 16, 2012

We had a small conversation on that and I read Dion's post.
While I disagree with some statements like

But systems of record are increasingly 1) becoming commoditized by SaaS and the cloud and 2) most organizations have reached the carrying capacity of the approach: There’s very little left to store and automate that isn’t already

I do adhere to the idea that we're settling into the infrastructural layer all the things that we take for granted: the simple, static stuff that doesn't change, is highly structured, rigid even. The data, the smallest building blocks, bits and bytes. Order reigns here, where everything is subjected to business rules, and automation can thrive

What is left and floats on top, is what is closest to humans: I said it 15 years ago and will keep repeating it, but humans will always stay on top of the food chain.
It's where the opposite takes place: Complex, dynamic stuff that changes all the time, unstructured, flexible itself and requiring the greatest flexibility at the same time. It's where knowledge and information flows freely, uncapturable. Chaos thrives here, exceptions are the rule, and automation usually is impossible, and when it seems to be so, that often turns out to be a very dear fata morgana

Systems of Engagement - that would be a good name for that all, but regardless of that, it's the very place where they would reside. Dion gives a few examples of SoE (just making up the abbreviations as we go here) but I'm not too fond of "online communities, crowdsourcing, Social CRM, open APIs" as SoE's, although they're Dion's invention of course, and it's hard to automate what can't be automated, like I showed

But maybe I just have to sleep on it and, as the hardcore business application guy I am, who thinks anything can get automated, everywhere, and effectively and cost-efficent even: have I ever envisioned what the future of automation is when indeed most is automated, and what remains simply can't be automated?
No.
But I know what it will look like. It goes something like this:

As discussed above and in my freeBook, Social is all about people-stuff, and comes with all the mentioned characterisations and attributes. And that is far, far away from infra. So can it be automated even? Well no, hardly. It's a bit like the cop standing on a highway, redirecting you to different lanes because one or two have been closed down due to an accident. Yes it's a rule, but is it worth automating? No. Never. Or hardly ever

So, back to John. PaaS? Salesforce? To "do" Systems of Engagement? There couldn't be a worse place. Yes you can PaaS your communities and run your default bulletin boards on them, but if you really want to fence them off you need good... customisation! And that you do close to the customer, local, with the vendor or an SI.
While I personally (hope to) see an end to both their ways, there'll be a few Winters and Springs before any bark will set sail in that direction. At the top of the foodchain customisation will reside, as always, and it will be lifted up from down below by SaaS driving the tertiairy and maybe secondary business processes to a company. The primary business processes are by nature what distinguishes them from the competition, and will never be SaaS-ed

Having said all that, it seems that Dion's Systems of Engagement will reside in the same place where a company's primary processes will be automated: the top-notch stuff will be developed in house or maybe outsourced, but certainly not will they be fit for SaaS

PaaS? If you think PaaS is a good place for your Systems of Engagement, why don't you just hand over your money to me? I promise I'll spend it all in one night at a bar, and that will ROI more than what you had in mind