5/11 - meet me on / quaternion

This post was written in prison.

[Warning. You are exiting stroboscopic space. Time and matter are no longer discrete.]

This will be my last one before I get out! See you out there. Once again I ran out of characters (& jail time). Quick addition: I'm not surprised or upset algorithmic stablecoins have bit the dust as an idea.

Book Reviews, Briefly & Finally

• The Quest for Artificial Intelligence - Nilsson

  4.0/5.0 (calculated with a three-layer feed-forward perceptron) - A pleasant survey on the history of AI. Written in 2010 by a legend, doesn't pretend to predict the future or get into too much technical detail. Well-documented/referenced.

• Meta Math! - Chaitin

  3.5/5.0 - Suffers from what most pop math books do. Still good brain food.

• Code Breaker - Isaacson

  3.5/5.0 - Something about Isaacson pisses me off. I'm not sure what it is. I admire how much he rolled up his sleeves on this one, though. A good story, if not a bit premature to document. Isaacson seems to want to cover history in the distant past (Einstein, Leonardo) as well as historic moments as they are happening (Doudna, Musk).

• A Random Walk Down Wall Street - Malkiel

  3.5/5.0 - A reasonable read. Very rosy assumptions on American equity exceptionalism but generally right on the difficulty (futility) of generating alpha.

• Einstein - Isaacson

  3.0/5.0 - There are dozens/hundreds of books on Einstein. Why?

• Three Body Problem - Liu

  5.0/5.0 - Literary perfection. Transcends the genre. Can't wait for the Netflix special.

• A Mathematician's Apology - Hardy

  2.5/5.0 - Writers write, mathematicians prove. Not sure why this is canon to anyone.

• Artemis - Weir

  3.0/5.0 - Quick, fun & dumb. 

• Superintelligence - Bostrom

  4.0/5.0 - Thought-provoking work on AI's future in the ballpark of Aaronson.

• The Storyteller - Grohl

  2.0/5.0 - Maybe I just like being contrarian, but this is rubbish. Self-indulgent, no warts or even scars from a prominent rocker. Just a book on, gee whiz, what a great guy the author is.

---

Personal

I'll be home quite soon. Two important questions: 1. How is my cat doing? 2. Will I be unbanned from Twitter?

If you like my writing now, wait until I'm reunited with my large language models!

---

Futures (I always believe in)

I ran out of room in the last letter and was having fun synthesizing some wild predictions of what the future will look like. One might wonder what benefit could possibly be derived from such an effort. I think it's important to explore the set of possibilities that could arise before they do. We know what the world looks like now, but we have a hard time enumerating the other set. It's inherently unpredictable, guaranteed to surprise us in every way. But, if you try, you may be able to not just define some collection, but more importantly, determine if there exists a map relating them. Is the latter set "onto"?; borrowing from algebra, perhaps you can think about them as one-to-one or bijective, as well. We search for functions that will map present conditions (say, in 1977, public key/asymmetric cryptography) to futures (distributed ledgers). What took so long to go from A to B? 

  Isn't it funny that in every case it seems to me that a new innovation to 'put down roots' before a prerequisite can build on top of it. We have to kick it around and decide whether we like it or discard it. A recent podcast interview with Carroll on Quanta's phenomenal series made a similar suggestion for QM--listen if you have the time! (I read the transcript, thanks Dad!)

---

Computing/Technology & Society

2025 - Autonomous vehicles emerge as the successor to mobile computing. A trend towards spending 8+ hours a day in your car begins. TSLA nears $10T market cap, Musk becomes the first trillionaire.

2026 - An AI-generated (GPT-4?) song, with zero human editing or direct participation, charts on Billboard.

2030 - Virtual reality/metaverse (a la snowcrash/RP1) is ubiquitous and widely used. A comfortable, ergonomic optical interface was the gating item.

2030 - Autonomous vehicles reach majority penetration in the US; vehicular accidents drop more than 50%.

2035 - An AI-generated book "charts" on NYT Bestseller list. Zero human involvement, including pesky editors.

2035 - AI humanoid robots are ubiquitous (a la Amazon's Astro but bipedal & thoughtful/near AGI).

2040 - Reinforcement learning humanoid robot soccer team wins World Cup-level match (or any other physical sport).

2040 - Autonomous vehicles (new or upgraded ICE) are mandated by law; vehicular accidents drop by 95% from prior era.

2050 - Lifespan for a newborn is 100-125 in the US. 90% of non-surgical medicine is automated. A non-insurance digital health/HCIT company ranks in the top 5-10 healthcare companies by market cap. Most of contract law is automated or "on the chain", anyway.

2051 - Travel time reductions have improved logistics, productivity & leisure: hyperloop, eVTOL, ultrasonic & stratospheric travel begin an aggressive adoption curve.

2055 - AGI humanoids are granted equal rights, marriages between humans & AGI are common.

2060 - An AI-generated movie wins best film. Holographic-style images are indistinguishable from reality.

2065 - # of Americans who live with companion animals (I don't like that "pet" word!) are on a -10% year/year annual decline

2070 - AI sentences a felon. Value of metaverse assets equals meatspace assets, like-for-like. For example, securities of companies belong to neither. Through another lens: metaverse-specific GDP is 50-150% of meatspace-specific GDP.

2080 - Lifespan for a newborn is 150-200+ in the US; 99% of all of medicine, criminal justice, politics & law is automated. Most carbon-based humans have not worked in decades. Money has generally lost meaning and value.

2100 - Immortality is expected within humans timeframes. AGIs vastly outnumber carbon-based life. Happiness is not judged to be any higher than 100 years ago.

Socioanthropology, Religion, Philosophy, & Politics

2022-2040 - Fall of Russia, Belarus, North Korea, Syria & Iran as we know it. The last dictators.

2025 - Supreme Court strips federal government of additional rights. Americans are freer than ever.

2025 (alternate) - Congress outlaws Bitcoin or crypto private wallets. Scenarios ranging from protests (guaranteed) to mass immigration (possible), to secession (unlikely) to civil internal armed conflict (very unlikely, but possible) are in play.

2028 - Possible first LGBT or woman, or both, becomes POTUS.

2030-2040 - Distributed Autonomous Government - a sovereign entity uses DAO/Blockchain technology to canvas the populace at the least, and more likely, to govern or legislate.

2030-2040 - A relatively large war is fought either nearly or completely autonomously.

2032 - Possible first LGBT or woman, or both, becomes POTUS.

2035 - Animal rights expand substantially.

2035 - Substantial decline in US university/college enrollment.

2040-2050 - Universal Basic Income in the US - results in unanticipated, unexpected collateral events.

2040-2050 - A major Western state is insolvent - let's hope not the US?!

2040-2050 - Blockchain referendum or elected official vote in the US; surveillance state efforts are repelled by growing libertarian base, despite social justice movement maintaining momentum. A strange merger of both ideologies coalesces and a third party is viable.

2045 - Marijuana is legalized in the US. State-based lotteries (such as Powerball) are outlawed.

2050 - Major sports (including NFL, NBA, MLB, NHL, EU Football) viewing is down 50% from peak, some leagues are disbanded or even banned due to lack of interest or health concerns; competitive boxing events are outlawed.

2050-2060 - An AI is elected as the leader of a sovereign entity, possibly a major Western democracy; two party system in the US has collapsed, may be retained in name.

2050 - majority of US high school/18-year-olds do not attend university/college

2060-2070 - Simulation hypothesis (Bostrom-complete) represents the largest segment, or even the majority of the US populace religious/meaning-of-life belief system.

Math/Complexity

2025-2070 - Riemann, Goldbach, infinite Mersenne proofs

2030-2055 - P != NP proven, continuum hypothesis proven false

2030-2055 (alternate) - P != NP proven undecidable

2030-2060 - Substantial improvement in theorem proving software, from DeepMind or OpenAI results in unexpected results

2030-2060 - Substantial improvement/simplification of Wiles-Fermat - maybe not enough to fit a small margin, though

2050-2100 - major non-QC breakthrough in integer factorization, enough to change cryptography standards

Quantum Computing

2035 - 1 second lossless qubit

2040 - 1,000,000 qubit computer

2045 - Halting problem oracle

2050 - many-body molecular simulations (1-5 seconds) routine

2060 - molecular simulations nearing macroscopic amounts powered by AI pruning algorithm are routine and contribute significantly to materials science, biomolecular research

Electronics/Engineering/Hardware/Physics

2030 - Tokamak-based fusion begins. Computing is 20% of global electricity usage.

2040 - Human spaceflight to Mars

2045 - Either Elon Musk, or Jeff Bezos, or both announce they are permanently relocating to space for the purpose of exploration (no return trip planned)

2050 - Mars or Moon human settlement; Space flight is routine, ISS expands, space/orbiting satellite hotels have occupancy of >50,000

2050 - Global carbon neutrality, synthetic carbon removal/sinks (not trees) emerging; climate engineering relevance/discussion peaks 5-10 years earlier

2050+ - memristors ubiquitous

2050 - EVs reach 1000 mile range with routine weight due to density/efficiency improvements

2050 - room-temperature superconducting ubiquitous

2050+ - Tokamak-based fusion accounts for >30% of global energy generation; cold fusion proven impossible with AI prover

2060+ - 0.1-1.0 petahertz chip/Moore's law hard limit reached for silicon. Adiabatic computing.

2060+ - Laser-accelerated interstellar probe at 0.5c

2060+ - wireless energy transmission - download electricity credits to your iPhone. iColoumb? maybe iWatt

2065 - quantization of gravity discovered

2065 - spooky action proven artifactual, c confirmed including information

2065 (alternate reality) - c can be violated under certain presently unclear circumstances

2100 - extra-solar system community with >1m population

Unsorted: smart materials, swarm robotics, high-throughput xenorobotics, DNA origami, computer-designed organisms, living cell engineering

Money/Finance

2025 - quantitative finance 'exploits' become more common after a major quant fund melts down due to malicious market-based action

2030 - Crypto > global liquid currency, same-day trade settlement for equities in the US

2040 - Crypto > global M1

2050 - More than half of global assets are tokenized/on-chain

2055 - A DAO is the largest company/financial entity in the world

2060 - JPMorgan or Bank of America is acquired by a cryptocurrency-oriented company

Biology/Medicine/Chemistry

2023 - PROTACs approved, begin to become an important medicinal strategy, eventually reaching 20%+ of all new approved medicines

2028 - Biosimilars are fully interchangeable by default, significantly reducing global drug spend.

2030 - First effective treatment for Duchenne Muscular Dystrophy approved. Liquid biopsy has saved thousands of lives.

2031 - Generic ASOs, ADCs, cell therapies & mRNA emerge.

2035 - First effective treatment for Huntington's Disease approved.

2040 - CRISPR broadly utilized for various cell-specific (non-liver) diseases. Liquid biopsy has saved millions of lives.

2041 - First disease-modifying drug for non-hereditary Alzheimer's approved.

2045 - HIV cure via CRISPR. Type 1 diabetes substantial treatment/cure without immunosuppression. Xenotransplantation gains share

2050 - Targeted senescence/fibrosis treatment approved

2055 - Personal health operating system reports real-time CBC, tissue-level genetic sequencing via SaaS application.

2060 - Germline editing is common. AI robotic surgery is routine. Liquid biopsy monitoring is near-continuous for some patients.

2065 - Incidental metastatic cancer is rare. Cancer has conceptually shifted to a quantitatively deterministic assay of clonal fidelity/instability.

2070 - General senescence/fibrosis treatments widely used, approved on surrogate markers and AI results.

Computing

2022: web3 begins, 5m users are browsing decentralized sites

2025: web3 has 100m DAUs

2030: 1 million transmon qubit quantum computer. Web3 500m DAUs

2030: post peta-flop era

2030: Most software is written by AI.

2040: AGI emerges, FSIQ ~100.

2045: There are more AI beings than humans, FSIQ reaches ~130.

2050: "Teleportation" of small molecules.

2060: Dr. Evil paradox empirically tested.

2065: DNA computing on trajectory to rival silicon.