Personally I'm not at all worried about the "hobby". I think it will not only survive but get more interesting. And I really believe the greatest achievements in modeling are still ahead.
I got started as a 5 y/o with the traditional Lionel set around the Christmas tree on Christmas morning. This was in 1961. My set was that US Army version with the blue loco and rocket launchers. It was very formative.
After retiring from the US Army ( I said it was very formative) I had too much time on my hands. So I went back for another college degree, this one in mechanical engineering. I'm 53 now. All four of my kids have participated to one degree or another in model RR.
Model railroading's present incarnation as a specialized hedge fund benefiting a few insiders is doomed. This is not an outcome I'm worried about, either. I don't have any dogs in that fight. From the viewpoint of my 9 y/o that "hobby" is counterproductive at worst and irrelevant at best.
We know how this hobby "developed" starting in the later 1960s. The brass loco importers paved the way. They were basically labor arbitragers offshoring manufacture to Japan, later to Taiwan and now to mainland China, buying lots of ad space in the prestige hobby press and pocketing the difference. This and demographics produced a financial bubble not that different from the late real estate bubble.
It became a vast exercise in model con$umeri$m. Buy manufactured models, then make custom scenery to display these models.
"Want to keep the hobby going? Find young folks and inspire them."
Many young folks today are in about the same situation as young people were in the early 1930s. They have far more time, energy and imagination than $$$. This is not a demographic that appeals to the shrinking prestige model RR press and its advertisers. Meanwhile Walmart and the schools already spend trillions training kids how to be consumers of imported plastic. They don't need any extra practice with Model RR Con$umeri$m. To be clear they can't afford very much. The real incomes of families with 10-18 year olds have been stagnant or declining for a decade. What you think depends on whose statistics you believe.
To the extent it has one I think the future of this hobby has to be (re) built around learning to make things again. And making things using updated tools and techniques. Tools like the personal computer, CAD programs and do-it-yourself CNC machines (Computer Numerically Controlled). Look at "Make" magazine. Watch "Mythbusters" and several other "builder" programs on the tube. Look at any of 100,000 instructional techy vids on Youtube.
One of the first model railroading "books" I got was the 1950s "Practical Model Railroading", first published by Kalmbach in 1952. This included a picture of the layout at Bell Labs. It's been awhile since model railroading was associated with that kind of technological leading edge.
People complain about R/C flying "stealing" all the kids. I suggest going into an R/C hobby shop. You'll notice very few ready to run aircraft are offered. What you primarily see is a store full of "parts" and "tools" to use in "building" r/c aircraft. Unsurprisingly, small low cost CNC machines have been far more quickly adopted by R/C hobbyists.
The typical model railroad hobby shop now is a near mirror image. The balance of ready to run and diy building is reversed. This is sad when you know that one of the first DIY CNC lathe articles was written by an electrical engineer who was also a model railroader. He adapted an old Atlas Press 6" lathe to automated computer control (CNC) so he could mass produce the wheels he needed for his narrow gauge model RR project. The article appeared in a very low circulation pub, I think Model Engineer. Incidentally, far more complicated circuit projects were presented by MR as far back as the late 1970s.
This particular lathe article was clearly not something Atlas (the model RR company), Athearn or Walthers would favor Kalmbach or Carstens running. If they did that what would be next? A four part series on designing and machining plastic injection dies to produce a custom run of model parts in a small home built die injection molder? Plastic vacuum formers?
http://www.lindsaybks.com/dgjp/djgbk/inject/index.html
http://www.lindsaybks.com/dgjp/djgbk/injatt/index.html
http://www.lindsaybks.com/dgjp/djgbk/vacf/index.html
Dozens of plans are readily available for 3-axis CNC machines carrying Dremels, Rotozips and trim routers around 2' work tables. The total cost to build is about $300 or so including stepper motors and driver boards.
http://www.engadget.com/2006/06/29/how-to-build-your-own-cnc-machine-part-1/
If you want to really dumpster dive. Junked laser printers and copy machines are the best source of stronger stepper and servo motors. And what's that? Acid etching and drilling circuit boards? No! You're supposed to BUY DCC controls from Digitraxx. Everyone knows this.
http://www.solsylva.com/
Lots of things can be done with this. Including routing your own HOn2 milled roadbed sections for Maine Two Footers, if you're so inclined.
http://www.crankorgan.com/plans.htm
The "Jester" was specifically designed for pocketing aluminum blanks for plastic die injection.
Other realities intrude here. For instance, once a hobby acquires a large base of small CNC lathes, CNC routers and CNC mills, and hobbyists with enough expertise in CAD and CAM, what then? Dies, molds and near instant hobby micro-manufacturers are close behind.
The model railroad manufacturers and distributors (a/k/a "advertisers") have always resisted articles that spill too many beans. Unimats were long advertised in those small corner ads. And it was possible to read MR and RMC for years without really learning what this obscure European device could do. Model Rectifier Corporation's entire existence is predicated on keeping hobbyists low skilled, rather than doing it themselves as generations before in the 1920s - 1950s did. And as they could do now.
To relate this to real life. Even in the current dismal economy CNC machine tool programmers can readily find work. The master key to this field is COMMAND OF CAD PROGRAMMING. Really learning to draw with CAD takes some time. It's why a lot of excellent manual machinists can't cross over to CNC.
Consider a hobby defined in such a way that a young man spends six-eight years doing CAD designs and then making parts from these designs with hobby level CNC machines. This "hobbyist" will have very solid CAD and CNC skills by age 18.
The same depth of "roll your own" is possible on the electronics side.
It's pointless to blame Kalmbach, Carstens or the other publishers for the present state of affairs and focus on railroad model consumerism. They have fixed overhead to pay. They followed a specific consumerist publishing business model. Selling advertising to businesses who sell to the reader-consumer is a critical part of this plan. This was decided on the day A.C. Kalmbach printed his first run and then went to mail it to hobby dealers.
But the internet wasn't online that day...
And in the case of Model RR these advertisers are almost entirely manufacturers and distributors of manufactured models. They disfavor articles that could turn readers from consumers into non-consumers and even into competitors. Ditto for articles that will redirect cash flow towards Mouser, Digikey, stepper motor vendors, Enco, etc. Commercial publishers and editors cannot go to perma-war with their advertising bases.
This set the stage for model railroading's Age of Scenery. John Allen, Bill McClanahan, Dave Frary, Malcolm Furlow, George Selios... All fantastic masters btw. Build spectacular scenery on which to display your manufactured model collection.
I do think blame can be generously distributed to the leaderships of the NMRA for the last two decades. They had far less excuse for allowing themselves to be co-opted into extreme consumerism by the manufacturers and publishers.
"perhaps inspiring my daughter or nephew or other kids and adults into pursuing, even if in passing, a neat part of my life."
There's a lot more to model railroading than pure tech. But if learning useful life skills don't reenter the equation in a big way then I do think the "hobby" will sink into very deep obscurity.