Of all the forms of tech debt, decoherence is the worst. It compounds exponentially with a system's rate of change, and AI is now producing it faster than ever, leading us deeper and deeper into the abyss...
Genuinely interesting article. I have just one remark:
resulting in runaway exponential growth
I’m always wary when posts throw around “exponential growth”, or expressions like “99% of times”.
Is it meant literally, or is it just a figure of speech. The latter would be disappointing for an article with a scientific slant, especially since it repeats it 4 times.
I ask because, honestly, I’m not convinced that the source of decoherence described in the post could produce anything faster than polynomial growth.
The exponential growth part is meant literally, I genuinely believe the growth in complexity of truly decoherent systems in combinatorial and exponential. It can be extremely bad!
The biggest issue with proving this in a meaningful sense is that it would rely on a lot of counterfactuals. The best example that I am familiar with is probably various layers and functions of US government bureaucracy. You can find many descriptions and examples of how layered and complex government systems are (see for instance Recoding America), and I would categorize this as decoherence. My sense is that if we made these systems more coherent, then they would also be much more efficient and much easier to change. But we can’t know for a fact that this would be true, because it would essentially be an alternate history.
On the other side though, I think we can compare the store of Windows vs Linux. At one point, Windows Server was the dominant server implementation and growing. Linux was the underdog and it was not treated seriously, with significantly less funding put toward it. Yet in the last 20 years, Linux has become absolutely dominant by count of machines spread across the world, and Windows really only maintains its foothold on the desktop.
Now, imagine if Linux hadn’t been successful, and specifically, imagine building Docker exclusively for Windows. Imagine having to manage not only the complexity of spinning up and tearing down Windows servers dynamically, but also the complexity of the licensing, the various bugs and issues you could encounter, and being dependent on Microsoft entirely for fixes since it would all be closed source. That would, IMO, be a decoherent mess, and the economic impact would likely have been far fewer successful startups, many orders of magnitude less economic activity, and a much less vibrant web.
Genuinely interesting article. I have just one remark:
I’m always wary when posts throw around “exponential growth”, or expressions like “99% of times”.
Is it meant literally, or is it just a figure of speech. The latter would be disappointing for an article with a scientific slant, especially since it repeats it 4 times.
I ask because, honestly, I’m not convinced that the source of decoherence described in the post could produce anything faster than polynomial growth.
score 1 · canonical
The exponential growth part is meant literally, I genuinely believe the growth in complexity of truly decoherent systems in combinatorial and exponential. It can be extremely bad!
The biggest issue with proving this in a meaningful sense is that it would rely on a lot of counterfactuals. The best example that I am familiar with is probably various layers and functions of US government bureaucracy. You can find many descriptions and examples of how layered and complex government systems are (see for instance Recoding America), and I would categorize this as decoherence. My sense is that if we made these systems more coherent, then they would also be much more efficient and much easier to change. But we can’t know for a fact that this would be true, because it would essentially be an alternate history.
On the other side though, I think we can compare the store of Windows vs Linux. At one point, Windows Server was the dominant server implementation and growing. Linux was the underdog and it was not treated seriously, with significantly less funding put toward it. Yet in the last 20 years, Linux has become absolutely dominant by count of machines spread across the world, and Windows really only maintains its foothold on the desktop.
Now, imagine if Linux hadn’t been successful, and specifically, imagine building Docker exclusively for Windows. Imagine having to manage not only the complexity of spinning up and tearing down Windows servers dynamically, but also the complexity of the licensing, the various bugs and issues you could encounter, and being dependent on Microsoft entirely for fixes since it would all be closed source. That would, IMO, be a decoherent mess, and the economic impact would likely have been far fewer successful startups, many orders of magnitude less economic activity, and a much less vibrant web.
score 1 · canonical