Two Campers and a Tiger
There is a joke that I believe is applicable to so many scenarios in real life that it deserved a blog post. It involves a scary animal (usually a lion or a tiger, but sometimes a bear) and two campers. This webpage contains the only written version of this joke I could find online. It’s an old HTTP site – I’m pasting that page’s version below so that you don’t have to click that link:
Two men are walking through a forest. Suddenly, they see a tiger in the distance, running towards them. They turn and start running away. But then one of them stops, takes some running shoes from his bag, and starts putting them on. “What are you doing?” says the other man. “Do you think you will run faster than the tiger with those?” “I don’t have to run faster than the tiger,” he says. “I just have to run faster than you.”
Yes, what that camper did was probably not the loveliest thing to do. I’ll be skipping the discussion on whether the behavior of the camper is ethical or not. Now, let’s consider the real life scenarios where this joke is applicable. My main observation is that people frequently confuse which person/thing corresponds to which character in the joke above. The following table indicates what the incorrect and correct mappings are:
| Lion | Sneaky camper | Other camper | |
|---|---|---|---|
| Incorrect | must be achieved/reached | new/proposed way/solution | [ignored] |
| Actual situation | ideal/unachievable scenario | new/proposed way/solution | current way/solution |
The story of the lion and the two campers tells us that new ideas/things don’t need to be perfect to be useful – in most cases, they just need to be better than the things they want to replace.
I’ll be going through a few examples.
Example scenarios
I have personally come accross all these scenarios. One motivation to prepare this post was to link this article whenever I come across another similar scenario.
Covid-19 deaths even with vaccines
Person X: “There were many deaths due to Covid-19 despite all those vaccines. Then what was the point of vaccinations?”
Person Y: It would have been a miracle to have a vaccine that saves everyone [running faster than the lion]. The vaccines saved many lives that would have been lost if there weren’t any vaccines [getting vaccinated gets mapped to putting your running shoes on]."
ChatGPT’s correct answer rate vs. DuckDuckGo’s correct answer rate
Person X: “ChatGPT occasionally gives me wrong answers. Why would I use it?”
Person Y: “DuckDuckGo (or Google, alas) also occasionally gives you wrong answers. So, we should instead ask ourselves whether ChatGPT performs better than DuckDuckGo. If it does, why would we not switch to ChatGPT?”
AI vs human doctors
Person X: “AI sometimes misdiagnoses or misses, so we should not rely on it. In fact, a recent study has shown that an AI model has missed $p$ percent of cases of disease $D$.”
Person Y: “It is correct that AI is currently not mature enough to be relied on completely. However, humans can also misdiagnose or miss diseases. What percentage of cases of disease $D$ were missed by humans?”
Person X: I don’t know.
Person Y: Then we cannot reach a conclusion.
Bitcoin vs USD illicit transaction rate
Person X: “$p$ percent of all Bitcoin/ETH/etc. transactions are connected to illicit activities. Isn’t that ridiculous? We should ban Bitcoin/ETH/etc.”
Person Y: “Before Bitcoin’s illicit transaction rate is declared to be good or bad, it must be compared to the illicit transaction rate for EUR/USD/CNY. Do you know those rates?
Person X: ?? No.
Person Y: The story of the tiger and the two campers!
Cryptocurrency vs TradFi energy consumption
Person X: “Omg, the Bitcoin network consumes so much energy. Isn’t that ridiculous? We should ban Bitcoin/ETH/etc.”
Person Y: “The Bitcoin network does consume (and waste) so much energy. However, for a fair comparison, we should also know the energy consumption of the industries that it is trying to supplement/replace (probably very hard to measure/estimate, though). Do you know that number?
Person X: ?? No.
Person Y: The story of the tiger and the two campers! Also, check out the Merge of Ethereum and see how much it reduced the Ethereum network’s energy consumption.
Using privacy-enhancing tools
Person X: Sure, I can use Signal and avoid Google Search, but I will still be tracked online in other ways. Then what’s the point?
Person Y: You don’t need to achieve full privacy for the life changes you make to be worth something.
ChatGPT!
As with many other things, ChatGPT had something to contribute to the discussion. It turns out that a fallacy called nirvana fallacy captures the issue I’m describing in this article:
Me: Let’s say that an artificial intelligence-based program was invented that can make diagnoses by scanning MRI scans, which is what radiologists do. This program has a higher accuracy rate than humans, but it is not perfect. It sometimes issues wrong diagnoses. However, the fact that it has a higher accuracy than humans leads it to replace many radiologists. If someone argues that this program should not be used because it occasionally makes mistakes, would they be committing a fallacy? If yes, what is the name of this fallacy?
ChatGPT: Yes, if someone argues against using the AI-based program solely based on the fact that it occasionally makes mistakes despite having a higher accuracy rate than humans, they would be committing a fallacy. The fallacy at play here is known as the fallacy of perfection or the nirvana fallacy.
The fallacy of perfection occurs when someone rejects a proposed solution or course of action because it is not flawless or does not meet an unattainable standard of perfection. In this case, the argument against using the AI program is based on the premise that it occasionally makes mistakes, disregarding the fact that it still outperforms human radiologists in terms of accuracy.
While the program may not be infallible, its higher accuracy rate suggests that it can contribute significantly to diagnosing medical conditions more reliably than human radiologists. By focusing on the occasional mistakes while ignoring the overall improved performance compared to humans, the person committing the fallacy is failing to recognize the value and benefits provided by the AI program.
It’s essential to evaluate technological advancements and interventions based on their overall effectiveness and benefits, considering their potential to improve outcomes, even if they are not perfect or free from occasional errors.