there is an inefficiency here
I agree 100%.
I think there are several sources of inefficiency in trying to find what you need in the doc. Listing those may help in building solutions.
First, 9 has become a fairly large system, no longer just the Radian engine with an SQL interface. So covering all it can do necessarily involves a lot of text to be searched to find what you want. As it gets built out we will increasingly encounter network effects between new features, where adding what may seem at first glance to be a simple new feature might geometrically expand what can be done given existing capabilities. 9 could get very much larger, requiring much more documentation, general education, and tips.
Second, 9 is evolving fairly rapidly. It often does so in internal/infrastructure ways that have subtle, but important, effects on the user interface. That has the virtue of trying out new things in a community driven model, but the downside of relearning in the context of frequently changing documentation. It could be that a topic already read needs to be re-read to work the thing as it is today.
Third, 9 is still adding user interfaces that provide greater point-and-click ease, with many more to go, which would reduce the need for more complex learning, such as learning to use SQL. For a core constituency of data engineers the use of SQL is, in many ways, self-documenting, in that knowing it and having a list of functions with a few examples gets you off and running. But there seems little point in teaching SQL in point and click style when, before such teaching can be written, point and click dialogs that are easier to learn will appear.
Last, I think is the limitation on documentation resources. Given finite resources the first task is to document what is there, to get it all down in an understandable English explanation. That comes first, and then, given more time, can appearsummaries and quick guides for various specific interests. Manual indexing is like that too, because it is very costly to change as topics are constantly altered.
To give you an idea of the resources required, I think in the past three weeks there have been several thousand changes involving nearly 400 topics, with over a thousand illustrations being added or changed. Changes are barely incorporated by the time the next build arrives.
I think in many ways there is a fundamental trade off: for the lowest cost and greatest capability, it helps to have a system that is evolving in a no frills way. You have to be fearless in buying into improved engineering, stronger machinery, and better GUI, no matter what the cost to existing documentation.
Once the main parts of the GUI are filled out in what is likely to be enduring form, then I expect we'll see improvements like quick "how to" guides, better indexing, jumps to topics from within dialogs, and so on.
Until then, we can use this forum as a big, parallelized, search engine, collectively using our intelligence and our individual exploration of release notes, experimentation, and documentation, to help each other, whenever any one of us gets stuck.