This won’t scale

I’m a software engineer. One concept that constantly comes up in software engineering is "scaling". We have to make sure any systems we write are going to be functional over the long term with large amounts of data. How large? Try to imagine how much data a hospital in a major metropolitan area could accumulate over the course of a decade… Yeah we’re talking a ton of data.

If you’re the paranoid conspiracy theory type, try to imagine how much data a government could accumulate on a population over the course of a month if it were really trying. Again, a lot of data.

As such a common criticism of certain systems, especially among the NoSQL crowd when talking about SQL and relational databases, is "this won’t scale" or "this doesn’t scale". When that is said, it means that the system works well on small systems and small amounts of data, but up on the level of Google won’t work well at all.

And when you’re talking about revamping the entirety of the United States health care system to be mostly if not entirely single-payer universal, you need to think on the scale of Google.

This is why I’m continually puzzled when leftists and other supporters of Obamacare point to Sweden as the example Obamacare can live by. Okay let’s look at this.

According to Wikipedia, Sweden’s national population is 9,555,893. That’s just north of 9.5 million. The population of the United States is estimated to be north of 316 million. In other words the United States has more than 33 times the population of Sweden. That’s like comparing your personal financial management system with that of a medium-sized corporation in terms of the number of transactions that have to be managed.

And apparently in Sweden, universal health care is working quite well. But that doesn’t mean that universal health care will work well on the scale of the United States. On the scale of New York City, it might, since the population of the City of New York (2012 estimate: 8,336,697) is slightly less than the entire Kingdom of Sweden.

For something that works on a small scale to also work on the large scale, adjustments must be made. Okay massive adjustments must be made. To get an idea of the kind of adjustments that need to be made, see if you can talk Quicken’s engineers into telling you the functional and systemic differences between what they offer for enterprises (i.e. large corporations on the scale of Google and Microsoft) versus small and medium businesses versus personal and home business use.

But there’s a complication: something that works on the large scale doesn’t work on the small scale. It’s a paradox those of us in the software development field know all too well: adjustments and design decisions made in a system to get it work on a large scale can cripple it on the small scale. In other words, if we could get universal health care to work in the United States, it would not serve as a viable example for Sweden.

Sweden has also had universal health care for decades, and their population over the last 10 years grew by less than 1 million people. As such they’ve had plenty of time to adjust and scale their system to keep it working over the long run by making small adjustments here and there. But the system they have now likely wouldn’t go over well 20 years ago, probably also not even 10 years go.

That is what happens when you try to scale a system up. The problem is that the other universal health care systems around don’t work well on the larger scales, so the United States does not have a good model we can use for a starting point, which is why the United States is trying to make it up as they go along, starting with Obamacare. Problems abound with the health care systems in Canada (2011 census: 33,476,688) and the United Kingdom (2011 census: 63,181,775).

And when you’re talking about scaling up a health care system, especially a universal one, population is what matters.

So perhaps we need to talk to Google to get an idea of how we might be able to scale up universal healthcare from the size of Sweden to the size of China and still have it work. Except, there’s another caveat: not all systems can be scaled up.

This is another concept that escapes a lot of people, not just politicians when talking about healthcare, but software engineers when talking about designing systems and software. There are inherent limitations in every system, and an upper boundary by which it can be scaled up. Google alone accounts for 1% of the entire world’s power consumption. Few companies can get that large. Most never will.

As such universal health care will face increasing costs with increasing scale, and the correlation is not linear. Eventually it will reach a point where it will be too expensive for the economy to handle. Sweden will never reach the size of the United States, so we cannot use them and their small-scale universal health care system as any kind of guide for how the Untied States should model anything.

Instead, health care needs to be managed at the State level at its highest point and the Federal government needs to back out.