Discover Top Posts Tagged with #complexity

Every complex ecosystem has parasites

I'm on a 20+ city book tour for my new novel PICKS AND SHOVELS. Catch me at NEW ZEALAND'S UNITY BOOKS in AUCKLAND on May 2, and in WELLINGTON on May 3. More tour dates (Pittsburgh, PDX, London, Manchester) here.

Patrick "patio11" McKenzie is a fantastic explainer, the kind of person who breaks topics down in ways that stay with you, and creep into your understanding of other subjects, too. Take his 2022 essay, "The optimal amount of fraud is non-zero":

https://www.bitsaboutmoney.com/archive/optimal-amount-of-fraud/

It's a very well-argued piece, and here's the nut of it:

The marginal return of permitting fraud against you is plausibly greater than zero, and therefore, you should welcome greater than zero fraud.

In other words, if you allow some fraud, you will also allow through a lot of non-fraudulent business that would otherwise trip your fraud meter. Or, put it another way, the only way to prevent all fraud is to chase away a large proportion of your customers, whose transactions are in some way abnormal or unexpected.

Another great explainer is Bruce Schneier, the security expert. In the wake of 9/11, lots of pundits (and senior government officials) ran around saying, "No price is too high to prevent another terrorist attack on our aviation system." Schneier had a foolproof way of shutting these fools up: "Fine, just ground all civilian aircraft, forever." Turns out, there is a price that's too high to pay for preventing air-terrorism.

Latent in these two statements is the idea that the most secure systems are simple, and while simplicity is a fine goal to strive for, we should always keep in mind the maxim attributed to Einstein, "Everything should be made as simple as possible, but not simpler." That is to say, some things are just complicated.

20 years ago, my friend Kathryn Myronuk and I were talking about the spam wars, which were raging at the time. The spam wars were caused by the complexity of email: as a protocol (rather than a product), email is heterogenuous. There are lots of different kinds of email servers and clients, and many different ways of creating and rendering an email. All this flexibility makes email really popular, and it also means that users have a wide variety of use-cases for it. As a result, identifying spam is really hard. There's no reliable automated way of telling whether an email is spam or not – you can't just block a given server, or anyone using a kind of server software, or email client. You can't choose words or phrases to block and only block spam.

Many solutions were proposed to this at the height of the spam wars, and they all sucked, because they all assumed that the way the proposer used email was somehow typical, thus we could safely build a system to block things that were very different from this "typical" use and not catch too many dolphins in our tuna nets:

https://craphound.com/spamsolutions.txt

So Kathryn and I were talking about this, and she said, "Yeah, all complex ecosystems have parasites." I was thunderstruck. The phrase entered my head and never left. I even gave a major speech with that title later that year, at the O'Reilly Emerging Technology Conference:

https://craphound.com/complexecosystems.txt

Truly, a certain degree of undesirable activity is the inevitable price you pay once you make something general purpose, generative, and open. Open systems – like the web, or email – succeed because they are so adaptable, which means that all kinds of different people with different needs find ways to make use of them. The undesirable activity in open systems is, well, undesirable, and it's valid and useful to try to minimize it. But minimization isn't the same as elimination. "The optimal amount of fraud is non-zero," because "everything should be made as simple as possible, but not simpler." Complexity is generative, but "all complex ecosystems have parasites."

America is a complex system. It has, for example, a Social Security apparatus that has to serve more than 65 million people. By definition, a cohort of 65 million people will experience 65 one-in-a-million outliers every day. Social Security has to accommodate 65 million variations on the (surprisingly complicated) concept of a "street address":

https://gist.github.com/almereyda/85fa289bfc668777fe3619298bbf0886

It will have to cope with 65 million variations on the absolutely, maddeningly complicated idea of a "name":

https://www.kalzumeus.com/2010/06/17/falsehoods-programmers-believe-about-names/

In cybernetics, we say that a means of regulating a system must be capable of representing as many states as the system itself – that is, if you're building a control box for a thing with five functions, the box needs at least five different settings:

http://pespmc1.vub.ac.be/REQVAR.html

So when we're talking about managing something as complicated as Social Security, we need to build a Social Security Administration that is just as complicated. Anything that complicated is gonna have parasites – once you make something capable of managing the glorious higgeldy piggeldy that is the human experience of names, dates of birth, and addresses, you will necessarily create exploitable failure modes that bad actors can use to steal Social Security. You can build good fraud detection systems (as the SSA has), and you can investigate fraud (as the SSA does), and you can keep this to a manageable number – in the case of the SSA, that number is well below one percent:

https://www.congress.gov/crs_external_products/IF/PDF/IF12948/IF12948.2.pdf

But if you want to reduce Social Security fraud from "a fraction of one percent" to "zero percent," you can either expend a gigantic amount of money (far more than you're losing to fraud) to get a little closer to zero – or you can make Social Security far simpler. For example, you could simply declare that anyone whose life and work history can't fit in a simple database schema is not eligible for Social Security, kick tens of millions of people off the SSI rolls, and cause them to lose their homes and starve on the streets. This isn't merely cruel, it's also very, very expensive, since homelessness costs the system far more than Social Security. The optimum amount of fraud is non-zero.

Conservatives hate complexity. That's why the Trump administration banned all research grants for proposals that contained the word "systemic" (as a person with so-far-local cancer, I sure worry about what happens when and if my lymphoma become systemic). I once described the conservative yearning for "simpler times," as a desire to be a child again. After all, the thing that made your childhood "simpler" wasn't that the world was less complicated – it's that your parents managed that complexity and shielded you from it. There's always been partner abuse, divorce, gender minorities, mental illness, disability, racial discrimination, geopolitical crises, refugees, and class struggle. The only people who don't have to deal with this stuff are (lucky) children.

Complexity is an unavoidable attribute of all complicated processes. Evolution is complicated, so it produces complexity. It's convenient to think about a simplified model of genes in which individual genes produce specific traits, but it turns out genes all influence each other, are influenced in turn by epigenetics, and that developmental factors play a critical role in our outcomes. From eye-color to gender, evolution produces spectra, not binaries. It's ineluctably (and rather gloriously) complicated.

The conservative project to insist that things can be neatly categorized – animal or plant, man or woman, planet or comet – tries to take graceful bimodal curves and simplify them into a few simple straight lines – one or zero (except even the values of the miniature transistors on your computer's many chips are never at "one" or "zero" – they're "one-ish" and "mostly zero").

Like Social Security, fraud in the immigration system is a negligible rounding error. The US immigration system is a baroque, ramified, many-tendriled thing (I have the receipts from the immigration lawyers who helped me get a US visa, a green card, and citizenship to prove it). It is already so overweighted with pitfalls and traps for the unwary that a good immigration lawyer might send you to apply for a visa with 600 pages of documentation (the most I ever presented) just to make sure that every possible requirement is met:

https://www.flickr.com/photos/doctorow/2242342898/in/photolist-zp6PxJ-4q9Aqs-2nVHTZK-2pFKHyf

After my decades of experience with the US immigration system, I am prepared to say that the system is now at a stage where it is experiencing sharply diminishing returns from its anti-fraud systems. The cost of administering all this complexity is high, and the marginal amount of fraud caught by any new hoop the system gins up for migrants to jump through will round to zero.

Which poses a problem for Trump and trumpists: having whipped up a national panic about out of control immigration and open borders, the only way to make the system better at catching the infinitesimal amount of fraud it currently endures is to make the rules simpler, through the blunt-force tactic of simply excluding people who should be allowed in the country. For example, you could ban college kids planning to spend the summer in the US on the grounds that they didn't book all their hotels in advance, because they're planning to go from city to city and wing it:

https://www.newsweek.com/germany-tourists-deported-hotel-maria-lepere-charlotte-pohl-hawaii-2062046

Or you could ban the only research scientist in the world who knows how to interpret the results of the most promising new cancer imaging technology because a border guard was confused about the frog embryos she was transporting (she's been locked up for two months now):

https://www.msn.com/en-us/health/other/horrified-harvard-scientists-ice-arrest-leaves-cancer-researchers-scrambling/ar-AA1DlUt8

Of course, the US has long operated a policy of "anything that confuses a border guard is grounds for being refused entry" but the Trump administration has turned the odd, rare outrage into business-as-usual.

But they can lock up or turn away as many people as they want, and they still won't get the amount of fraud to zero. The US is a complicated place. People have complicated reasons for entering the USA – work, family reunion, leisure, research, study, and more. The only immigration system that doesn't leak a little at the seams is an immigration system that is so simple that it has no seams – a toy immigration system for a trivial country in which so little is going on that everything is going on.

The only garden without weeds is a monoculture under a dome. The only email system without spam is a closed system managed by one company that only allows a carefully vetted cluster of subscribers to communicate with one another. The only species with just two genders is one wherein members who fit somewhere else on the spectrum are banished or killed, a charnel process that never ends because there are always newborns that are outside of the first sigma of the two peaks in the bimodal distribution.

A living system – a real country – is complicated. It's a system, where people do things you'll never understand for perfectly good reasons (and vice versa). To accommodate all that complexity, we need complex systems, and all complex ecosystems have parasites. Yes, you can burn the rainforest to the ground and planting monocrops in straight rows, but then what you have is a farm, not a forest, vulnerable to pests and plagues and fire and flood. Complex systems have parasites, sure, but complex systems are resilient. The optimal level of fraud is never zero, because a system that has been simplified to the point where no fraud can take place within it is a system that is so trivial and brittle as to be useless.

If you'd like an essay-formatted version of this post to read or share, here's a link to it on pluralistic.net, my surveillance-free, ad-free, tracker-free blog:

https://pluralistic.net/2025/04/24/hermit-kingdom/#simpler-times

#complexity

Trending Tags

Recently Viewed Tags

#complexity