squidproquo

When OpenAI goes public that will be the sign that they need to pass the baton to the retail invest--I mean--bag-holders.

The non-determinism is part of the allure of these systems -- they operate like slot machines in a casino. The dopamine hit of getting an output that appears intelligent and the variable rewards keeps us coming back. We down-weight and ignore the bad outputs. I'm not saying these systems aren't useful to a degree, but one should understand the statistical implications on how we are collectively perceiving their usefulness.

Reminds me of the marketing for a really bad movie:

- “genius” - “the missing link” - “breaks the barrier”

The other thing to note is that journalism in the US has gotten really lazy. A lot of the articles you will see in the MSM are based on leaked info and press-releases from PR firms, etc. It's easier to for journalists to regurgitate stories hand-fed to them than doing truly hard and costly investigative work.

Agree with this. Human language is also not very information-dense; there is a lot of redundancy and uninformative repetition of words.

I also wonder about the compounding effects of luck and survivorship bias when using these systems. If you model a series of interactions with these systems probabilistically, as a series of failure/success modes, then you are bound to get a sub-population of users (of LLM/LLRMs) that will undoubtedly have “fantastic” results. This sub-population will then espouse and promote the merits of the system. There is clearly something positive these models do, but how much of the “success” is just luck.

Thank you for this. It’s nice to feel not alone.

Excellent blog post. Summed up what I've been thinking for the past few years. I find myself endlessly consuming content (videos, podcasts, etc.) that I would deem "educational". It feels good, like you're learning something new, improving yourself. But at the end-of-the day you aren't. Your brain is just sitting there passively - not really doing any hard work. Instead of your brain generating ideas and solving problems, that process has been palmed off to the internet.

A lot of these instances seem to describe meteorites. I don't see a connection to lightning at all. Also, it's pretty crazy that the ball lightning phenomenon has been around for a while, yet it hasn't been replicated in a lab.

I'm pretty sure you can get Knorr Aromat Seasoning in Europe. It's not pure MSG, but it does contain it (probably the main ingredient).

Interesting, I built an app that does something similar with respect to converting forecasts to trades, allowing forecasters to trade their predictions (in hidden market) for points (not real money).

https://www.unitarity.com/app/

You will probably score very high on this test: https://www.psychopathtest.com/

An interesting PBS Space Time episode covering this: https://www.youtube.com/watch?v=Rh898Yr5YZ8

This site is crowdsourcing forecasts for the location of flight MH370: https://www.unitarity.com/app/challenges/mh370/events/2020

If anyone is interested in forecasting the location of where flight MH370 will be found: https://www.unitarity.com/app/challenges/mh370/events/2020

I built this project a few years ago with a Django back-end and and Angular as a SPA front-end. The component that shows the 3-D Earth was built with d3.

The goal of the project (Unitarity) is to make crowdsourcing forecasts for various event outcomes easier and more intuitive. Coming from a background in finance and tech, I always had this unsettled feeling that so much time, energy, talent and money is spent in the world of finance to win a basically zero-sum game. It's hard to quantify the positive externalities that are generated from this game, but if I had to guess, they wouldn't be that much. Think of all the stories of Physics PhDs who leave academia to use their skills to work in finance. At some point, the more energy you put into a zero-sum game, there is a net loss of productivity. What if we could redirect those skills to a game that has positive externalities and benefits society?

The site was initially seeded with a few challenges to demo the different types of events and forecasting scenarios. For more info on the workings check out the guide (https://www.unitarity.com/app/docs/guide).

When I first started the project, I wrestled with what frameworks to use. As a Rails dev, I really wanted to use Ruby for the back-end. However, Python is more familiar to those in academia and data science. I ended up choosing Django (which I really like in a lot of ways). IMHO, it has better (and intuitive) support for DB migrations than Rails. On the downside Django, has really bad documentation and the community of 3rd-party modules is not as big as the Rails one. I really disliked writing tests in Python compared to Ruby. I really got used to RSpec and came to like it quite a bit, so testing in Python wasn't so fun.

Before I jumped into using a JavaScript front-end, I tried simple html templates. However, Django is not as good as Rails when it comes to templating. In Rails, one has a plethora of formats to choose from (erb, haml, etc.), that Django is missing. SPAs were all the rage around the time I started coding and since I had some experience with Angular, that's what I went with. The app has quite a bit of UI interactions so going with an SPA made a lot of sense. Angular gets a lot of bad press, I found it to have an intuitive framework that uses dependency injection (it was familiar to me coming from a Java/Spring background). Also, Typescript was a real joy to use. I originally started as a Java developer 20-odd-years-ago, so having a typed language meant less testing to make sure things were working.

Overall, it was a fun and challenging project to work on, and I learned a lot. However, I still wrestle with some of the technical decisions. For example, whether-or-not I should have built it as an SPA, since this was quite a learning curve and had some unforeseen consequences. SPAs are a helluva lot more complicated than just building a bunch of html templates. In many ways, doing an SPA can make you feel like it's Groundhog Day. You are basically building a parallel front-end application that can be more complicated than the back-end. The SPA will have many of the same models and routes that the back-end does and coding it will sometimes feel like you are going through a boring routine of repetitive tasks.

There's more I can go into, but I wanted to keep this brief.

This is not a great investment. It's pays it's base rate (0.0%) plus inflation (7.12%). The rate is based on inflation and is reset every 6 months, and I'm assuming the base rate of (0.0%) doesn't change. It may make sense for some people who are risk-averse and already have savings in a bank account that is getting demolished by low rates and high inflation.

I think the overarching idea here makes sense, but the devil is in the details.

1) It's going to be difficult to ensure that the 'Orb' device can't be hacked and allow bad actors to send a bunch of random codes to signup for a Worldcoin.

2) If there are bogus signups how do you deal with reconciliation when there are ID collisions?

3) It's going to take many years for this to get a critical traction and to deploy coins to individuals.

4) The value of this coin will be extremely small when it's divided amongst the world's population. Assuming it's spent shortly after it is received, this money will just flow back to the owner's of capital anyways.

This article reminded me about a strange incident that happened to me a decade ago. I was with a group of friends in a bar playing a game of Trivial Pursuit. None of us were drunk or under the influence of drugs. Sometime, during the middle of the game, it was my turn to answer a question. The question was something like "What is the name of the city in Central America whose name translates to 'silver hills' in the native Nahuatl language?" I sat there dumbfounded, having no clue as to what the answer was. I was about to give up. But, before I could, I heard a voice in my head slowly whispering something. At first I couldn't make the words out, but as the voiced repeated itself, it became clearer. I blurted out: "Tegucigalpa!" The correct answer. My friends looked at me a little bit shocked wondering how I could have know that. The answer must have been buried deep in my mind somewhere, for it to come to me through a voice whispering the answer to me was strange an a little bit unnerving. Nothing like that has happened to me since.

As reassuring as when Palau joined the "Coalition of The Willing."

If anyone dabbles in forecasting the Consumer Price Index (CPI), add your forecast here: https://www.unitarity.com/app/challenges/us-consumer-price-i...

Star Trek replicator technology v0.0.0.0.0.1

Reducing the number of hours/jobs that people currently work would reduce the labor supply and wages would rise (Econ 101). I don't see how these "consequences" are a bad thing.

Furthermore, "unforeseen consequences" are effects that we don't know will happen because they lie beyond our ability to predict them. There will always be unforeseen consequences of any major policy decision.

If anyone has been doing Covid-19 simulations for the United States. I'm crowdsourcing forecasts on this site: https://www.unitarity.com/app/challenges/us-coronavirus-outb...

Interesting article. In late March I built a simple SEIRD model for a few US states with python. It was very difficult to estimate R0. R0 is essentially Beta/Gamma, where Beta^-1 is is time between contacts and Gamma^-1 is the time until recovery. I think what should be added to this analysis is population density. Each state has a different rate that people come into contact with one another. It would make sense that R0 in NYC is much higher than in Iowa.

Also this website is crowdsourcing US COVID-19 mortality forecasts if anyone is into data modeling:

https://www.unitarity.com/app/challenges/us-coronavirus-outb...

This website

I ran some SEIRD epidemiological simulations (I'm not an expert nor in that field), but the death rate in the NY region was indicating a R0 between 4 and 6, not the 2.5 or so I've seen in a bunch of places. Obviously, population density is a factor in that number as well. I'm trying to aggregate forecasts for the US on this website if anyone is interested: https://www.unitarity.com/app/challenges/us-coronavirus-outb...

If anyone has a forecast for the United States, this website is aggregating forecasts:

https://www.unitarity.com/app/challenges/us-coronavirus-outb...

After living in London for a year and coming back to the US, I was perplexed at how much better (and cheaper) the UK supermarkets were when it came to prepackaged food. I rarely ate at restaurants and did a lot of cooking. Sainsbury's had a really great selection of pre-made chicken and tuna salads. They cost like $2 US (equiv) and all you needed to do was get some bread and it was enough to make two sandwiches. In the US, similar items cost around $6 (so over-priced). I would go to Pret quite a bit too, but there was also another chain that was cheaper (can't remember the name), but after like 4PM all the sandwiches they didn't sell got discounted and you could get some good ones for half-price.

Here is a link to a news article that less technical: https://www.newsweek.com/milky-way-dark-matter-kepler-145323...

What I don't understand is how these "studies" seem to find things that would require a baseline to compare with. Results like, brain damage was suffered, cognitive impairment, brain-shrinkage, all would require the subjects to have some baseline snapshot before the "sonic attacks" occurred.

Cool idea!