Posts
-
Review: How To Invent Everything
CommentsHow to Invent Everything is a book by Ryan North, of Dinosaur Comics fame. Its premise - time travel is real! You’ve just used your time machine, but now it’s broken and you’re stuck in the past. You want to:
- Not die.
- Recreate your modern life of luxury.
How to Invent Everything doesn’t literally cover everything, but you can see it as a book about the highlights of civilization. It covers basic survival tips, building up to important technologies like the compass, then ending with fundamental science concepts and a basic intro to computing and logic gates. Its main thesis is that with the right reference text (itself), you could reproduce 12000 years of human progress in, say, 5 years of work.
Odds you’re ever in the exact same scenario as the premise? Vanishingly small. Still, it’s a fun read if you’re interested in exploring how humanity improved its quality of life from first principles.
* * *
The main reason I like this book a lot is that it’s very, very reductionist. That necessarily falls from the premise, but it makes it easy to trace the links between technological innovation. Steam engines relied on precision engineering to make airtight pistons. That required a standard unit of measure and knowing how to make steel. Steel requires knowing how to smelt iron, and that requires kilns. To get the kiln to high enough temperature, you want it to be made out of bricks, and bricks are easy if you know how to make charcoal. There are many details to fill in for getting the raw materials, and the book deliberately focuses on the technology rather than the politics of convincing people to help create your vision, but having the road map to go from nothing to the core of the Industrial Revolution is a big deal.
(It’s also fun to consider how different things would be if certain natural resources weren’t so plentiful on Earth. So much technology traces its lineage to the chemical and physical reactions that turn wood and mud into useful materials, and it’s not like wood has to exist on a planet with life.)
Over and over again, the book points out the difference between when inventions were invented, and when they could have been invented. The gap is usually decades, or centuries. It’d be easy to conclude that humans are dumb, and in some sense that’s true. If you assume that evolution increases intelligence very slowly, and civilization appears as soon as an organism is smart enough to construct it, then the natural conclusion is that humanity is the dumbest species that’s just smart enough to create civilization.
So yes, we’re dumb. On the other hand, when you read the “solution” to different technologies, it can be hard to fault humanity. The prerequisites for the bicycle were around for a long time, but considering their design, even minor changes make the bicycle just a novelty instead of a more efficient means of transportation. (See this post by Jason Crawford for more analysis.)
It’s also easy to see how environmental factors can hamper innovation. Take crop rotation for instance. The core idea of crop rotation is that if you grow the same crop every year, and haven’t invented fertilizer yet, your crops will use up nitrogen in the soil. Next year’s crop is bad, the one after that is worse, and eventually the soil is no longer fertile. The easiest way to fix this is to let fields lie fallow, which means growing no crops to give time for the land to recover. Alternatively, you can plant legumes like clover, which replenish nitrogen thanks to their symbiotic relationship with nitrogen-fixing bacteria.
Now, imagine you are a generic human around 10000 BC. You’ve just discovered the basics of farming, and have reached an uneasy level of food security. Some bozo is saying you should leave a perfectly good field alone for long term sustainability. Are you going to listen to them? Keep in mind that you don’t know the chemistry, you have no way of learning the chemistry because the right tools don’t exist yet, crop experiments have a long delay to payoff, and you have limited experimentation budget because messing up crops means people go hungry. It’s easy to see why we don’t have evidence of two-field crop rotation until 6000 BC, and four-field crop rotation wasn’t invented until around the 17th century. Discover the correct causal links seems very hard, especially given the other confounding factors.
I don’t want to give humanity too much credit. Hot air balloons took a really long time to invent, given that baskets, controlled fire, and fabric all existed for centuries. Incubators for premature babies were not used until 1857. All that needed was a warm box, and their introduction reduced premature infant mortality by 28 percent. The Pelton wheel is a more efficient waterwheel from the 1870s, about 1400 years after the earliest known waterwheel. Its primary differences are using pressurized water and a better paddle shape. It really doesn’t feel like discovering those two should have taken 1400 years, but it did. It’s made worse when the book cites a source that Pelton was inspired by spraying a cow in the face.
I mean, cool story, but it’s a bit embarrassing no one thought about the physics earlier.
* * *
Does How to Invent Everything offer any greater truths about the universe? Well, maybe. Here are my takeaways.
Models don’t have to be correct to be useful. Many times, people had the wrong idea about why a technology worked, but that didn’t stop it from working. For thousands of years, no one knew where alcohol came from when brewing beer, but people brewed it anyways. Similarly, there is evidence the Egyptians knew that eating liver helped them see in the dark, but it would take many years to learn this was because livers are heavy in vitamin A.
Technology can be lost. Rongorongo is a set of glyphs found on Easter Island that might be an independent invention of writing, but it’s hard to tell because no one knows how to read it. In Rapa Nui culture, only the privileged few could be taught how to read and write, and after a smallpox epidemic went through the island, anyone literate was dead and the symbols were just that - symbols. As another example, the cure for scurvy was lost and rediscovered seven different times between 1400 and its final discovery in 1907. That’s crazy. You wouldn’t think “eat a fresh orange” would be so hard for a society to remember, and yet there it is. Arguably, this is an argument we should focus on better archiving of information.
Society advances when information spreads easily. One question How To Invent Everything toys with is what point in history would let you most influence the trajectory of humanity. It concludes that the answer is either introducing spoken language, or introducing writing. Spoken language allowed humanity to share complex concepts with each other, which let ideas live after death. Writing (and later, the printing press) allowed people to share those concepts further than their community. Even if an entire village disappeared, if their writing remained, their ideas could persist.
Both these technologies are so incredibly useful to society that it’s hard to see how progress was made before their invention, and they look a long time to invent. Anatomically modern humans are estimated at 200,000 years ago, spoken language is estimated at 50,000 years ago, and writing is estimated at around 3200 BC. Again, I don’t fault humanity for this. Getting a community to agree on the meanings of arbitrary sounds seems really hard when no one knows what language is. Getting that same group to agree on how those sounds map to written glyphs seems extra hard. It’s a horrible coordination problem.
If the engine of invention is powered by people sharing random ideas until good ones emerge, then I can’t help but wonder if the best inventions are ones that make sharing ideas easier. Social media’s gotten a beating ever since 2016, but I am coming around to the argument that social media increases the volume of both bad information and valuable information. In this view, you can argue for free speech both morally and practically. Morally, people should not have to self-censor. Practically, progress can come from ideas that sound heretical in the present day, and it can be hard to tell whether history will agree with our judgment.
Atomic gardening is insane. Not actually important, I just thought it was cool. Find a source of radiation. Put it in the center of a field. Plant a bunch of crops around the radiation source at varying distances. Hope you get useful mutations. You have basically no control over the results, it’s just random search in genetics space. It’s the kind of crazy idea that could only come from the 1950s, when atomic hype was high and fear about radiation was low. Some grapefruit varieties sold today trace their lineage to atomic gardening, and shooting X-rays at Penicillium molds led to a mutation that 5x-ed penicillin production. I suppose history has judged it as a net positive, but I almost feel personally offended that such a brute force method worked.
-
Six Years Later
CommentsSorta Insightful turns six years old today!
If you’re new, every year I do a meta-post about how my blogging went that year. I’ll be honest - I’ve had a lot less motivation to write this past year. There isn’t a single reason for this. It’s more like a few reasons compounding together.
More Time on Social Media
I haven’t checked how much time I’ve been spending on YouTube and Reddit, but it’s definitely rising. Recommendation algorithms keep getting better.
I know these systems work well on me, so I try to avoid adding social media streams. I don’t have a TikTok or Instagram or Snapchat, since I don’t want to add any many mindless scrolling into my life if I can help it. We’ll see how long that lasts. For better or worse, those are the communication mediums of today, and culture goes forward with or without you.
Puzzlehunt Obligations
This past year, I’ve been doing more work for puzzlehunts. It’s hard work, but it’s fun.
I see puzzle construction as trying to tell a story where you have minimal control over how the story plays out. You’re trying to convey a solve path, but if you make it too obvious then the solve is no longer interesting.
The more relevant part re: this blog is that puzzle constructing is time consuming. It’s like game development. You place the solver in a system, they’ll stretch it in ways you never thought of, and you just have to run around patching up all the holes that become apparent. It’s also a team project with a hard deadline where I can read comments from people who are looking forward to participating. In comparison, this blog is a personal affair, with no strict deadlines or hype trains. That means I’ve felt greater feelings of responsibility and urgency for puzzlehunt writing compared to blogging.
To put hard numbers on this, based on my time tracker, last year I spent about 100 hours on blogging and about 435 hours on puzzlehunt construction. It’s a pretty stark difference.
Interest in my Hobbies are Cooling Down
A number of my posts were me writing about something I was passionate about. I had a post about a My Little Pony x Doctor Who crossover, a post about a My Little Pony x Euclidean geometry crossover, and a post about Neopets. It made sense for me to write them at the time, but I’m less into MLP now, and I’m less into Neopets ever since the site redesign, and I’m just not into fandom as aggressively as I was in 2015 or 2016. I still like MLP and Neopets, and keep up-to-date on what happens in those spaces, but they’re a smaller part of my life.
So when I think about writing a grand post explaining why I got into My Little Pony, it’s a lot harder to do so. When I consider starting a post about why you should read Gunnerkrigg Court, it’s hard because I think about Gunnerkrigg a lot less. There’s always Dominion, but at this point I’m basically retired.
What I’ve realized is that most of my posts are not long-burning posts. They’re works of passion - I bash them out in a few days, or I don’t finish them at all. Sometimes, that passion is a function of the time, rather than anything intrinsic about myself. I told a few people I was going to write an “offline RL manifesto” back in 2018, since I thought it was criminally understudied, but it’s since become a reasonably sized research area and that manifesto no longer feels as important. The post I wanted to write about measurement feels less important now that fairness / interpretability / AI safety is more widely accepted.
I know I just said blogging is a personal activity, but I guess one of my personal activities is writing up hot takes, and takes can’t be hot if you feel they’re already known. I somehow find it more rewarding to create a new point of discussion, rather than adding to or emphasizing an existing one.
I realize this isn’t how writing has to work. Preaching to the choir is fun, after all. However, it’s how I’m approaching writing right now. This is likely why I’ve found puzzle construction to be nice - puzzles deliberately aim to be novel.
Settling Into Patterns
I feel like more of my life has “normalized”, for lack of a better word. By this, I mean that if I picked some arbitrary week 1 month in the future, and made a prediction for what I would do that week, it’d be more accurate than if I did the same exercise a few years ago.
There’s just fewer surprises. I’m not meeting as many people, and I’m doing fewer new things. This is all quite terrifying, to be honest. I’m still figuring out what to do there, but seeing less means I’ve had less inspiration for blogging.
It doesn’t do you much good to draw unless you have something to draw, and the only place you’re getting anything to draw is out of that head. And the only way you can exercise the mind is by bringing new ideas to it so it’ll be surprised, and say “God I didn’t know that.” That’s the greatest thing in the world, that “Gee I didn’t know that.” And there you are, you know?
(Chuck Jones, director of several classic Looney Tunes cartoons. I recommend this Every Frame a Painting video, if you haven’t seen it before.)
I am spending more time on social media…but that doesn’t seem to drive much inspiration for me.
General Tiredness with COVID-19
I stopped blogging about COVID at some point, but I remember believing that things would go back to normal after vaccines were widely available. However, I didn’t account for the degree of vaccine hesitation, as well as the slower rollout of vaccines worldwide enabling evolution of variants that (might) escape vaccines developed so far.
My expectation is that things are and will get better, but last year, I could point to a clear future event (conclusion of widescale vaccine trials). Whereas this year, I don’t have anything to point to besides rising vaccination rates (yay!) and new variants (boo), with an unclear sense on how both will evolve. That’s sapped a lot of energy out of my life in general.
* * *
I’ve considered whether all this means I should stop blogging. As an exercise, I’ve tried imagining the world where I never update this blog again…and it feels really bad. I do care about blogging, so I’ll try to slice out more time towards working on it.
Statistics
Word Count
Last year, I wrote 32,161 words. This year, I wrote 26,955 words.
5812 2020-08-18-ai-timelines.markdown 1112 2020-08-18-five-years.markdown 1549 2020-12-30-car-co2.markdown 6597 2021-01-29-mh-2021.markdown 4249 2021-02-18-flash-games.markdown 5420 2021-04-07-grad-school-5years.markdown 1018 2021-05-23-dominion-temple.markdown 587 2021-05-23-world-can-change.markdown 611 2021-07-05-blog-ads.markdown 26955 totalI wrote 9 posts this year. Some were quite long - this year I averaged 3000 words per post, compared to about 2500 words per post last year.
View Counts
These are the view counts from August 18, 2020 to today, for the posts I’ve written this year.
12426 2020-08-18-ai-timelines.markdown 277 2020-08-18-five-years.markdown 1912 2020-12-30-car-co2.markdown 912 2021-01-29-mh-2021.markdown 167 2021-02-18-flash-games.markdown 2367 2021-04-07-grad-school-5years.markdown 102 2021-05-23-dominion-temple.markdown 210 2021-05-23-world-can-change.markdown 86 2021-07-05-blog-ads.markdownThis mostly tracks what I expected, although I’m surprised the post about Flash games has so few views. Maybe my nostalgia about Flash games is narrower than I thought.
Time Spent Writing
I spent 100 hours, 11 minutes writing for my blog this year, which is about 20 hours less than I spent last year.
That rounds to 16-17 minutes per day, which is actually not too bad for a time commitment (although in practice my blogging is very bursty, rather than writing a few minutes each day).
Posts in Limbo
Post about measurement:
Odds of writing this year: 20%
Odds of writing eventually: 30%If I don’t write this next year I’m going to remove it from my list of pending ideas.
Post about Gunnerkrigg Court:
Odds of writing this year: 45%
Odds of writing eventually: 50%Basically I am saying that if I don’t do it this year, I don’t think it will ever happen and will remove it from this list.
Post about My Little Pony:
Odds of writing this year: 50%
Odds of writing eventually: 90%I cannot see myself never writing about the My Little Pony fandom. It was a huge part of my life, but I’m ready to describe it as was, rather than is. Of course I say this when a lot of my music history is pony EDM…but outside of a few music artists I found via MLP, I’ve stopped caring about a lot of the brony fandom. Much of my enjoyment of MLP was derived from seeing how new episodes were interpreted by the fandom, and so far I have zero hype about Generation 5.
Post about Dominion Online:
Odds of writing this year: 25%
Odds of writing eventually: 50%Post about puzzlehunts:
Odds of writing this year: 70%
Odds of writing eventually: 90% -
Why Don't I Have Ads?
CommentsWhy doesn’t this blog have ads? Well, the answer’s simple: at my current view count, it isn’t worth it. And I don’t mean that it’s not worth the time to set it up. I mean that if someone offered to do it for free, I would still turn them down.
Let’s do a very rough Fermi estimation of how much revenue I could get. Google Analytics says I get about 5000 sessions per month. According to this blog, the clickthrough rate of display ads is 0.05%, but is 8.8x higher for native ads. I don’t want to do native ads, but let’s get an upper bound by assuming I did use native ads, and they have 10x clickthrough, so 0.5%. Average cost per click is $0.58. In total, the optimistic estimate is:
\[5000 \times 0.5\% \times \$0.58 = \$14.50\]Sure, that’s some money. However, ads also come at a cost to readership. Based on analysis by Gwern, running ads drops views by about 10%, and a drop of that size has been replicated independently across several organizations.
So really, the question is, would I rather earn at most $14.50 a month (likely much less), or would I rather have 500 more readers? Seems pretty clearly in favor of the readers side to me.
* * *
If finance has taught me anything, it’s that you can always take the inverse of a bet. If I value readers more than money, why don’t I pay money to run ads to direct people to my site?
At first, I thought it was dumb and stupid, but when I thought about it more I realized I didn’t have a good argument for why it was dumb, and I needed to take it seriously. I have free unclaimed AdWords credit, which Google offers to new accounts. Ads would increase viewership. They wouldn’t even affect the reader experience, because the ads would be on other people’s sites, not my own.
I think the main problem is that ads don’t really work for a personal blog. If your blog has a specific focus, you can advertise towards that specific form of content. My blog deliberately has no focus besides the things I want to write about, and I have no interest in changing that. There’s no clear search queries I’d want to sponsor. If someone searches my name, they find this site with ease (the benefits of having a novel last name), so it’s not like I need to pay for better placement. As for a banner ad, I don’t even know what image I would use. It’s like asking me to come up with a picture that summarizes my entire life.
There also could be follow-on effects, where someone views my blog differently if they clicked a sponsored link to get there. For example, I lost all trust in mattress recommendation websites after learning how online mattress advertising works. Most importantly, you can pay for views, but the rate isn’t great. At the previously mentioned price of $0.58 per click, a budget of $14.50/month only leads to 25 clicks per month. Much worse than the 500 views sacrificed if I did it the other way around! In retrospect, given how ads work, the price per click has to be based on the revenue of the products people advertise, and if I don’t sell anything, I’m priced out by everyone who does.
Given that I’ve never ever seen an ad for a personal blog, it probably doesn’t make sense to do so. Still, fun exercise to think through.