NEW YORK (TheStreet) -- Rather than adapt existing applications to the cloud, what's increasingly evident is developers are creating a cloud-based Internet layered on top of the existing Internet.Google ( GOOG), Amazon.com ( AMZN), Facebook ( FB) and all the other neat new services that deliver custom results to you in the blink of an eye are built on cloud. Pages are created on the fly by machines. They're not built by people and then pointed to, they don't have "Web addresses" in the way we think of them. They seem untouched by human hands. Whether you're searching for individual needles in a big haystack of data or seeking relationships within the haystack or trying to address huge numbers of mobile users within a small geography, the cloud world is materially different from the World Wide Web as first envisioned by Tim Berners-Lee. So how do you do "customer service" in this world? This, it turned out, was the topic of the "cloud chat" I was asked to do for IBM ( IBM) Thursday. It wasn't about cloud developers but about cloud customers, cloud users. What do you do when the cloud doesn't seem to have the droid you're looking for? I've had that problem myself, specifically at Amazon. Someone paid me roughly $30 last year for something they bought there, and I've been trying to track down what it was for tax purposes. This is where things break down. I've moved my email to the cloud because I no longer have time to filter out the spam, semi-spam and demi-spam filling my old PC inbox. Amazon didn't have the new address in its files, so its bots didn't know how to help me. Someone could have probably looked up the transaction record from the tax form sent to me but how do you find that someone? The technologies we use for customer service simply don't scale to cloud, because people don't scale to cloud. I tried email with Amazon but other companies have tried live operators, or live operators behind chat lines, and they've all failed. IBM itself sees some hope in tweeting, which was what our conference was really all about, but that seems just as unlikely to work.
All this gets serious when you move from buying a gift or tracking a business relationship to teaching and learning, which may be the most important thing we have to do today. I've written before about Moore's Law of Training, my point being that there is no such thing. We learn only as fast as we learn, and we teach only as fast as the student before us can absorb the subject. So how do you create a Massive Open Online Course, or MOOC, that has validity as a learning experience, no matter how good the teacher? What you find is that you have to build an immense human infrastructure behind the performance. People have to write quizzes, they have to write Frequently Asked Questions for each section, they have to organize virtual or in-person tutoring sessions, they have to find a way to go one-on-one with students who need extra help. The cost of all this dwarfs the actual cost of building and teaching the course. The same thing happens whenever you try to layer customer service onto a cloud application. You're trying to put running boards and training wheels on a Maserati. There are some private companies like Kana Software that see opportunity here, and claim to deliver "multichannel customer service," databases that can be used by clients and phone reps. But the same software also gets lost in the weeds of trying to analyze what people are saying about you on social networks, or analyzing customer interactions after they've happened, when it might be better to get things right in the first place. It's this tension between the instant gratification of cloud and the clunky interaction of real human beings that may be the biggest challenge facing the new Cloud World, and the biggest frustration for those of us back here on Planet Earth. At the time of publication the author had positions in IBM and GOOG. Follow @DanaBlankenhorn This article was written by an independent contributor, separate from TheStreet's regular news coverage.