Theme: Incentives

ALTERNATIVE HYPOTHESIS
Ryan (all),
We have been saying that engineering would take over product management for many decades. However, it can only happen if the innovations produced by engineering outpace the demands of the market AND are not in conflict with the demands of the market. So management will insert product management whenever those conditions fail.

Now, the rate at which we exhaust the market opportunity for any given innovation in technology is rather short – on the order of a few years to a decade. And as we become competent in the optimum means of exploiting any technological innovation, innovation decreases, and applications increase until the market is exhausted by a few surviving companies and technologies. Except in rare outliers, research and development on novelties is done by startups whose expenditures are “off book” of the larger companies that later acquire them and exhaust the market for their application. So the opportunity for the engineers to drive product design is largely in the early stage of the innovation, and that opportunity is exhausted by the success of the market.

Ryan (and Y’all) are young, and in tech especially, you keep trying to reinvent the wheel as if what you’re doing is novel, rather than repeating a cycle that occurs with every non-incremental leap in technological functionality.

So, using your example, information architects were necessary because engineering lagged innovations in user experience necessary to satisfy user and business demands. It wasn’t devs that solved it. It was training devs through iterative experience that the customer’s behavior not the code, drove the design of products. In other words when web interfaces were new, it only took a few years to standardize (exhaust) the limited vocabulary of design patterns. The same occurred on handheld devices. And we’ll see the same relationship between Product Managers and Devs as Information Architects and Devs. Because the original disequilibrium between the knowledge of the producers of the tech and the knowledge of the customers of its application is an equilibrating function over time.

And we will get there relatively quickly when we no longer have to guide the next generation of LLMS because they’ll understand our context and just start there every time. The only novelty with LLMs, is that we’d expected to have to go through the perception disambiguation, episode formation, prediction, valuation, and ethical sequence before we added the verbal interface. Instead, we’re working backward from the verbal interface because of the availability of information on the web, and the cost of compute reduced by what were previously graphics processors.

Y’all lack the historical knowledge (the fault of educators to understand the economics and operations of the industry rather than the craft of programming) to grasp the evolutionary process of operationalization (organizations) of innovation, using capital and people to satisfy markets over time. I’ve been writing software since ’74. And I can list off the top of my head the generational changes in software. It’s a very small number that doesn’t take the fingers of both hands. And much of what we’ve seen with the browser tech’s been bad other than the instantaneous distribution capacity for updates.

Instead, the primary problem with the industry is the rate of the expansion of the labor pool overwhelming the institutional and disciplinary knowledge base, causing this tendency for each generation to think they’re doing something novel instead of it’s finally economically possible given the hardware and infrastructure improvements to achieve what had been imagined by 1960.

AI is a great example. It’s just been made possible by a few geeks, that not that many years ago, figured out how to repurpose graphics cards that were invented for 3d video games for neural network data structures, thereby drastically improving the performance. And now we have AI processors on chips.

Otherwise, other than some shifts in the algorithms due to being able to actually test them on hardware, everything that’s happened since 1960 was pretty much deterministic.

And as I suggested for a decade now, the encryption age is about to die because there is no substitute for transaction processing, and two-part keys.

It’s very difficult to obtain knowledge of a field as an artist, craftsman, technician, or scientist. That’s a failure of our higher education system. They don’t teach the economic history of the field so they launch new generations of technicians into it who like the ‘non-sciences’ of psychology, sociology, and politics, lack the basic knowledge of physics, economics, innovation, and evolution. So those who are promoted do. And this creates a divide between craftsmen and management.

Anyway. It’ll happen as I’ve described above. Because that cycle is a fundamental law of physics, life, behavior, economics, and information. So expect the cycles.

Just how it is.

Cheers

Reply addressees: @rjs