... more on topic: Using flash to get the system installed typefaces is nifty, but a) doesn't seem to work consistently (some get rendered, others remain default styled) and b) you will use a looot of potential users just for using that technology.
The interface needs to be a bit more communicative. Use pointer cursors, selected states, and the flow of what to do when to what result is not as clear as it could be.
The relative weight of bolds and lights is also influenced by the use case. For example a bold geared towards screens use might be more explicitly dark to convey a proper weight increase on any number of resolutions and rendering algorithms, whereas a high quality editorial print might be a little more subtle and its weight increase can still be obvious enough. The other thing that also makes weight relative is the intended use size, and how cluttering up adds extra weight in small sizes, which you might compensate for to keep the weight distribution as intended.
But with both those issues it is a chicken and egg problem when it comes to designing the type.
Personally, I think this becomes a less absurd move when you think of brands not so much as static, monolithic entities, but as evolving actors that are eager to stay relevant and interact with their audience.
What Google has been doing with their own typefaces, and later with material design, is to visually own the entire market segment. Open sourcing "brand assets" is part of that proliferation strategy where your distributed presence drowns out competitors on a larger scale. In this context open source is just a modern, supposedly more candid way of trying to assume a market leader role (niche or major segment alike). Thinking back just a few years on early iOS and skeuomorphism, Apple dominated the smart phone market's design DNA by quite willingly escalating their own visual language down the entire ecosystem - it's a stretch, but consider something like their iOS HCI guidelines an open sourced design asset.
But even for comparatively smaller players this strategy holds up. Concrete Example: A consumer electronics retailer here in Finland uses Ubuntu Sans - the store gets the subliminal benefit of a typeface that evokes a techy look to your clueless average customer, yet at the same time Canonical (the company behind the Ubuntu operating system) is hardly lamenting the fact that it is their typeface that adorns a store which happens to also sells Windows and Macintosh PCs. While their motivation might have been rooted in the context of open source software more than aggressive marketing, it arguably is a branding win for them to have one of their open sourced brand assets being used in the context of consumer electronics that in itself goes beyond their own brand. And I dare same this exact same case works for IBM Plex, without it eating up the integrity of the main brand it originated from. If a tech blog uses IBM Plex, is detrimental or beneficial to the IBM brand?!
To come back my initial premise of brand relevancy the point I am trying to make is that the fact that we as a design community are discussing an aspect of IBM's branding is also contributing to their brand. Maybe two weeks ago you might have thought of IBM mostly as producing mainframes and working on government contracts, and maybe now you think of the company as somewhat more open to interaction with users and the design community at large. Even with open source that does not care about integrating back community feedback, the basic premise is that an actor gives to the public, free of charge. Charity has since forever and a day been something companies and brands are happy to associate themselves with. And more to the present day being, or even just appearing to be, design-driven is trendy - and so is open sourcing your typeface.