If you are going back to writing on stone (or clay tablets), you cannot ignore Delay line memory, including Mercury delay lines.
Delay lines were developed to store radar blips so that screens displayed only new, moving blips. In computers, delay lines converted data bits (ones and zeros) into sound waves, transmitted them acoustically, then converted them back into bits. They circulated forever until changed by the computer.
As well as complicating the architecture and programming of a computer using Mercury Acoustic Delay Lines, the mercury filled tubes were non standard components that had to be handled very carefully; they were bulky and had to be kept under close temperature control. Some feel of the mechanics of using them can be gained from looking at the description of the building of EDVAC, about 40% down the page (a chapter from a monograph on the history of Electronic Computers within the Ordnance Corps, written in 1961). EDVAC was the official U.S. successor to the ENIAC, with the design finalised in 1947; but it didn't start to work usably till late 1951 -- "by early 1952 it was averaging 15-20 hours of useful time per week".
Delay line memory was far less expensive and far more reliable per bit than flip-flops made from tubes, and yet far faster than a latching relay. It was used right into the late 1960s, notably on British commercial machines like the LEO I, Highgate Wood Telephone Exchange, and various Ferranti machines.
One more item to ad to your list: magnetic core memories. Back in the lates 60's, computers had magentic corememories, the cores strung amid an array of x-y wires that controlled its activation and reading response. similar designs were used in military aircraft at the time; all done on 1k and 2k of memory.
Cool! I didn't know that. I've just come to hate the marketers who thought to inflate spec numbers by using decimal rather than binary for their drive size measurement. You don't see them doing tht for RAM, just hard drives, from what I've seen. I'm always having to explain to friends and relatives why their hard drive did not format to the size they expected.
"A key parameter that isn't discussed is the likely life expectancy of the stored data. Clearly #1 (stone) wins hand down."
In addition to the life expentancy of the raw data, knowledge of the format is also important. This requires some continuity of knowledge. Even older forms of storage encounter a similar issue; writings in dead languages can be very difficult to understand.
(As you pointed out, even data that is physically readable may be uneconomical to read because of the cost of the reading device.)
My sense is that other than cars, almost nothing is improving in longevity and maintainability. Cars certainly last much longer than they used to and require much less maintenance. Almost all repairs, however, require a skilled mechanic. Even replacing a headlight on my wife's VW requires removing the windshield washer and annoying little repairs to the electrical components can be exceptionally expensive. Electronic devices like televisions have longer life expectancies than their tubed ancestors - but repairs are typically cost prohibitive.
I hadn't thought about that aspect. That is very interesting how that works. Maybe it's not just memory - things are getting easier and easier to manufactuure, but in some cases aren't built for nearly as long a life.
What are the engineering and design challenges in creating successful IoT devices? These devices are usually small, resource-constrained electronics designed to sense, collect, send, and/or interpret data. Some of the devices need to be smart enough to act upon data in real time, 24/7. Are the design challenges the same as with embedded systems, but with a little developer- and IT-skills added in? What do engineers need to know? Rick Merritt talks with two experts about the tools and best options for designing IoT devices in 2016. Specifically the guests will discuss sensors, security, and lessons from IoT deployments.