1. Everyone Is Lying Again: A blog dissecting how, in response to the majority of even slightly controversial or partisanship-evoking news stories, all “sides” (right, left, etc.) grossly misrepresent the facts and/or their scientific, historical, or cultural context. Probably one dissection per week, allowing time for substantial research for each post.
2. The Story of Rock Music: This podcast would guide the listener through the history of rock music, playing extended clips of >10 tracks per episode and helping the listener hear exactly how different styles developed, split off from their roots, and recombined later. For example in one episode I might explain what a raga is and play an example clip, then explain what modal vs. chordal improvisation is and play contrasting clips, then explain what blues rock is and play an example clip, then explain what post-bop jazz is and play an example clip, and finally talk through (with example clips) how these forms were fused together in the classic Mike Bloomfield track “East-West,” one of the earliest examples of “raga rock.” Episodes would proceed in roughly chronological order, so that the listener could “hear” the evolution of music over time, as later episodes build on the stylistic evolutions described in past episodes.
3. Evolving Sounds: Relatedly, I’ve long wanted to research, compose, and record a many-hour continuous piece of music that recapitulates the entire history of “Western music” (which is better documented than other traditions). The piece would begin with sections composed in accordance with scholarly guesses about how prehistoric music might have sounded, eventually transition into the earliest styles from recorded history, then evolve into styles covered in e.g. Burkholder’s History of Western Music, up to the present day. This is a pretty obvious idea and I’m upset that nobody has attempted it yet.
4. Everything is Awesome and We’re All Going to Die: This book would start off like a more thorough and epistemically scrupulous version of the empirical sections from Enlightenment Now, and would then proceed to explain in detail why global catastrophic risk is nevertheless increasing over time via Moore’s Law of Mad Science, inescapable asymmetry in the difficulty of creation vs. destruction, inadequate equilibria, and related phenomena.
5. Better Moral Judgments: Moral philosophy makes little attempt to estimate what our moral intuitions would be if we were smarter, better informed, etc. Works like The Righteous Mind and Moral Tribes ¹ are baby steps in the right direction, but still far less ambitious than what I sketch here, which could easily be expanded into quite a large research program. But I’d start with a book sketching out that research program and working through some initial examples.

Added on 9/10/2018:

6. The Case For and Against Utilitarianism: Among families of theories in normative ethics, I think the utilitarian family has the most to recommend it, but no book on utilitarianism I’ve seen covers (in much depth) what I think are the most important arguments. In part that’s because some of the best arguments build on a perspective derived from Better Moral Judgments (see above), which has yet to be articulated in any detail. But it’s also because books on utilitarianism often say very little about developments in utilitarianism after Mill (who died in 1868!) or maybe Sidgwick (d. 1900). Arguments I’d cover, in roughly descending order of how compelling I’d expect to find them after further investigation: (1) a variety of reasons why better-informed moral reasoners would trend toward utilitarianism (e.g. because utilitarianism is least parochial/arbitrary and simplest), (2) classic arguments for expected utility maximization + complete class theorems + Harsanyi, (3) utilitarianism handles personhood fictionalism better, (4) a variety of pragmatic considerations, such as the central thesis of Moral Tribes, and other things including “utilitarian theories are better specified,” and (5) the historical track record of utilitarianism. ² I’d also cover what I think are the best arguments against utilitarianism, e.g. that it might break on infinities (kind of a problem given that the universe is likely infinite). Compared to other books, I’d spend much less time on “Do my current, immediate, stupid and uninformed moral intuitions like or dislike what utilitarianism vs. other theories recommend in such-and-such highly constructed scenarios?”, for reasons that a copy of me in a different Everett branch has explained in Better Moral Judgments.
7. Moral Weight: Amazingly, in 2018, there is almost no literature attempting to estimate the comparative moral weight of different species’ lives and subjective experiences under varying assumptions. This book would build on some preliminary thoughts on the question I previously posted here.
8. Very Important Facts: A blog that, once per month, publishes a post explaining and making the case for some extremely important fact (that isn’t ~universally acknowledged), e.g. (a) “Almost everything we care about and can measure has gotten better around the world since approximately the industrial revolution” and (b) “Nuclear winter is plausible in the next 50 years and would be incredibly devastating” and (c) “The number of animals (farmed or in the wild) who probably/plausibly have lower subjective welfare than humans living in extreme poverty is X/Y/Z (under different assumptions), and we could probably substantially help A/B/C of them (under different assumptions) in reasonably cost-effective ways” and of course (d) “AGI will likely be the most important development in the 21st century.” In other words, the blog would only cover things that are massively more important (in a “how good/bad for how many?” sense) than the worldview/policy/culture/science debates that dominate the attention of most citizens/journalists/scientists/etc.
9. The Story of Film: Basically The Story of Rock Music (see above), but for film. Each video essay in the series would play >10 film clips per episode to show the viewer how different film techniques and styles evolved over time, split off from their roots, and recombined later. Episodes would proceed in roughly chronological order, so that the listener could watch the evolution of film technique and style over time, as later episodes build on the developments shown in past episodes.
10. Moloch: A TV series (I’d write the scripts) about coordination problems resulting in nuclear conflict and nuclear winter. The series would trace the stories of a variety of high-level decision-makers and influencers, as well as ordinary citizens living in multiple countries, as tensions between two great powers increase. By the end of the first season, a variety of coordination problems (hence the title) and instances of incompetence/carelessness (see e.g. The Doomsday Machine or Command and Control) would lead to a large-scale exchange of nuclear weapons. Subsequent seasons would focus on the surviving characters’ struggles to survive under conditions of nuclear fallout and extended nuclear winter (a bit like Threads). The series would strive for realism, broad accessibility, and emotional impact. My goal would be for the series to be watched by heads of state and other high-level policymakers, and impact their decision-making in ways that reduce the likelihood of nuclear conflict, as The Day After may have via its influence on Reagan and perhaps Gorbachev.

Added after 12/29/2018:

11. Alien planet BoTW/Sekiro game: A action/adventure game with the extreme freedom (including climbing and gliding) and nonlinearity of Breath of the Wild and the detailed and challenging combat+bosses of Sekiro. You play as a non-humanoid alien creature with >4 appendages and several strange but plausibly evolvable biological abilities, all of which manifest as fundamental gameplay mechanics. The game world is set entirely on an alien planet, allowing for varied and bizarre environments (which loop back into each other like Dark Souls), and varied and bizarre enemies and NPCs (none humanoid). As your skills increase, you gradually progress from parts of the alien planet that are primitive in technological development to areas with inhabitants at ever-greater levels of technological sophistication, until you finally discover a giant gleaming city with e.g. skyscrapers and space travel. ³ There is no dialogue; the story is told entirely via the world and your interaction with it, and the gameplay mechanics are discovered via “teaching without teaching” as in e.g. Shovel Knight. There is also no explicit leveling; instead, you get better only through practice plus the acquisition of new biological capabilities and equipment, few of which bear much resemblance to those acquired in games set on Earth.
12. Scaruffi album ratings predictor: The music critic whose aesthetic taste best predicts whether I will enjoy an album or not is Piero Scaruffi, who has provided 0-10 ratings for tens of thousands of albums across a wide variety of genres (but mostly rock and dance music). I wish somebody would train a model that can predict Scaruffi ratings for albums he hasn’t had time to rate, based on Scaruffi’s existing ratings, metadata from databases like RYM and AMG, the language of critic and listener reviews from a variety of platforms, ratings from other critics and listeners, etc. This is a pretty straightforward supervised learning problem, so presumably nearly all the work is in scraping and cleaning the data.

Footnotes:

1. I’d also list the methodology sections of Beckstead (2013) and these notes from a conversation with Carl Shulman.[↩]
2. I’m not that optimistic about the “historical track record” argument but it’d be interesting to explore. For examples of why I’m not that optimistic, see the “Philo” section in this blog post.[↩]
3. A twist at the end of the game is that when you climb aboard a space ship, you can launch into space. Decades pass, and you eventually crash-land in a giant body of water. The sky looks more Earth-like than anything you’ve seen on the alien planet throughout the game. Your space ship can function as a boat, and you eventually find a rocky shore with no discernible vegetation. Miraculously, your ship indicates the atmosphere is similar enough to that of your home planet that you don’t need a space suit, so you begin to walk inland. You eventually stumble on some creatures and their settlements. You have crash-landed off the coast of Germany circa the year 1300. Queue sequel.[↩]