Home

Predictions

Accountability and improving thinking skills.

Contents

• Background
• List
• See Also

Background

There are two reasons for this page.

First, prereigtration. After friend1 sent the group a link to I Can’t Believe Calling Him Hitler Didn’t Work, friend2 responded:

I wish pundits preregistered their election analyses like good scientists, by publishing holistic hot takes before election night, not after, and given partisan readers wouldn't want to read narratives that clashed with their preferences, the writers lock the hot takes behind a timer on a crypto chain and auto decrypts in 24 hours. This way, readers have much greater confidence that the analysis was well reasoned through and results in a thesis, rather than the other way around, where you start with the thesis (one side won and the other lost) and then backfill reasons for why, which becomes rather unfalsifiable.

This is my preregistration of results, and while I don't have an automatic crypto chain decrypter in place, Git does a good enough job. Dates, prediction (yes/no, probability), link to the question, and my reasoning are all listed below. I may not resolve each question out of laziness and forgetfulness.

Second, improving my thinking abilities and calibration. Answering these questions accurately requires asking plenty of others and answering those, creating a chain of question-answer pairs that uncover the entire issue and the ecosystem it resides in. Reading other predictions will also give me more ideas and questions to ask, further improving abilities.

Prediction are formatted on a 0-100% scale, with less than 50% being no and greater than 50% being yes. Two examples:

1. "No, 10%" means I predict a 10% chance of it resolving yes, so the likely resolution is no
2. "Yes, 70%" means I predict a 70% chance of it resolving yes, so the likely resolution is yes

Process vs. Outcome

A friend brought up a few important points regarding the differences between process and outcome in prediction (bullets paraphrased from his points):

• Reality must slap you in the face to realize your prediction was off
• Just focusing on the process is like philosophizing (sic) and never actually addresses any issues with your thinking

And I agree. My initial thought was that the process matters more than the outcome—it's developing the thinking skills, the ability to navigate the rabbit holes of questions that must be asked to uncover more of the map, that matters, else you won't get accurate predictions. But if I practice shooting a bow in the way that I think is best, but never actually see where my arrow lands, then it's pointless! I could spot-on and not know it, which would give me more confidence and conviction in my methods, or I could be completely off the mark, telling me that I review my ways of thinking and see where I lost it. Sure, the right questions must be asked, but the right answers must also be given, and the only way to check if the right answers were given is to check the score at the end of the test.

List

Template

• Date predicted:
• Prediction:
• Resolution:
• Link:
• Reasoning:

Will [person] go on [trip]?

• Date predicted: 06 January 2025
• Prediction: No, 40%
• Resolution:
• Reasoning:
  • Yes: Everyone went to Mexico and that was a pretty penny, but not nearly as much as what this would cost... maybe a $1000 difference? Most people are probably fairly well-off and could afford it, especially if it's a main trip for the year and known about this far out.
  • Yes: [person] is fairly confident that people would come and probably wouldn't be suggesting it if [person] didn't they would come.
  • Yes: Going on a three-day weekend helps those with minimal PTO.
  • No: Most people just don't go on international trips. This can be due to money, lack of time, lack of benefit, lack of interest, etc. Millenials and GenZ take about two international trips a year, so this slightly updates.
  • No: The cost increases slightly as more people don't go due to housing being more. The cost threshold may be slightly higher than the cost if everyone goes?

Will [person] buy [thing] by the next time I see them?

• Date predicted: 27 December 2024
• Prediction: No, 30%
• Resolution: True; yes, they bought it
• Reasoning:
  • Most people think that it won't happen to them and it's nothing to be concerned about. They also overestimate the severity of the event itself, and thus won't.
  • Things like this tend to be forgotten when you've been doing it the same way your entire life.

Will [person] save my phone number by the next time I see him?

• Date predicted: 23 December 2024
• Prediction: No, 35%
• Resolution:
• Reasoning:
  • Person and I normally communicate through Discord, but I got his phone number through a mutual friend to ask a question, and then texted a few days asking if he wanted to do an activity together. At said activity, I learned he hadn't saved my phone number. I viewed this as "little bro-ing" someone in some sense, and I think saving phone numbers is somewhat Lindy, i.e., if they haven't saved it after a week, it's unlikely they'll save it soon. Buttttt the better friends you are, the more likely they are to save it, so a bit conflicting.
  • Person laughed it off and didn't have a good explanation, which points to a no resolution.
  • Everyone I know who habitually don't save others' numbers just straight don't do it. This observation also points to a no resolution.

Will any healthcare company accept 10% more claims one year from now than they do currently as caused by the UH assassination and not another cause?

• Date predicted: 15 December 2024
• Prediction: No, 10%
• Resolution: [pending]
• Reasoning:
  • Increasing accepted claims as a result of the assassination would send a message that assassination works. Surprisingly enough, BCBS walked back a controversial anasthesia coverage change mere days afterwards, making me think that they are just waiting things to die down before reviving it. I would also not be surprised if a government entity encouraged them not to change out of fear that copycats may target other industries. There may be something here about the companies trying to appeal to the common citizen by saying "look! we've accepted way more people!", but I think it is just too dangerous to say that after an assassination.
  • Companies can simply beef up security, making future assassinations less likely. This includes obfuscating travel and engagement plans, having more physical security in the form of bodyguards, and removing high-profile executive images from media presences.
  • Increasing accepted claims would decrease profits ceteris paribus, resulting in dissatisfied shareholders and leading to a potential board turnover into one more aligned with shareholder interests. In other words, shareholders would just vote out a person(s) that wasn't trying to maximize profits and replace them with someone who was.
  • Healthcare companies have been viewed as a cancer on society for a long time (I do not necessarily endorse this view) and this will go back to being the normal status for them after all of this dies down. Like Kaczysnki, Mangioni may live on in infamy among a few devoted believers, but long-term impact on these massive companies is likely to be zero.
  • (As an aside for the "as caused by the UH assassination" qualifier, this can be quantified by looking at number of claims reviewed by AI, if there was a step function change in number of claim responses.)

Real-Time Translation Technology by 2028?

• Date predicted: 28 November 2024
• Prediction: Yes, 70%
• Resolution: [pending]
• Reasoning:
  • I suspect that improvements in translation quality is simply scaling compute and training data along with some basic RLHF, especially from native speakers. This is already accessible if companies want to focus on both, I'm just not sure if they are or will, leading into the next question...
  • How much of a market is there for real-time translation tech? I suspect high, assuming the quality is there. Businesses can interact without playing telephone via translator, making for more efficient and honest interactions; travelers can easily go to foreign places and feel more comfortable, again making for more efficient, enjoyable, and honest interactions with locals; friends and family members can better connect with each other. While English is and will almost certainly continue to be the world's lingua franca, the nuance (jokes, hidden meanings, historical context, idioms, etc) of each language often gets lost when translating to English.
  • There are a few types to keep in mind: voice-to-audio (listener translates speaker's language to target language), text-to-audio (reader translates text to target language audio), voice-to-text (speakers gets voice translated to text for them to write). Some are easier than others.
  • What would this look like? The tech would continuously listen to the speaker and translate as it saw fit to convey the meaning correctly and effectively. But this is just speech... what about visuals? Google Translate has the ability to recognize characters and translate on-screen. What if we could get some AR glasses to also recognize text in its field of vision, then replace those on the screen? This doesn't seem completely out of reach (granted I'm inexperienced with this kind of stuff, so I may be completely off the mark), although the overlap with the native text may be awkward. Could writing do the same thing? The speaker talks and the tech tells them what to write.

Bitcoin hits $100K before Trump’s inauguration?

• Date predicted: 07 November 2024 20:47 → 12 November 2024 16:40
• Prediction: No, 35% → yes, 60%
• Resolution: True
• Link: https://manifold.markets/predyx_markets/bitcoin-hits-100k-before-trumps-ina
• Reasoning:
  • Trump has been quite friendly to crypto in recent months. From Al Jazeera:

    The Trump campaign accepted donations in cryptocurrency and he also appeared at industry events, promising to make the US “the crypto capital of the planet”.

  • Trump isn't that great at keeping promises, so his crypto bro rally may have just been a ploy to get more voters. However, in my haste to go to bed when I made this prediction, I didn't consider that this doesn't really matter since his policies won't affect pre-inauguration... I guess I figured people would consider his lack of promises in their pricing. Or I'm just wrong!
  • The largest BTC gain was in August 2021 with a 17% gain. (I am wrong here again, as BTC went up 28% on the week as of the morning of November 12. I underestimated this!)
  • Elon Musk is likely to be a Trump advisor and he is very friendly towards crypto. This will encourage people to buy before the price increases. (Elon has been announced as a director of the new government agency, DOGE.)

See Also