Posts
Comments
I am wondering if there is any update on which AI research assistant you are settled with eventually and why?
Not really, because I spend some dedicated time to Ankify important stuff from the materials I have read throughout the day (or days) and this process involves a quick review and put them together.
Well said. "it's in any case quite useful to set effective priorities." this is exactly where I have been struggling.
Yes, I read through your annotated edition in detail as well and benefited a lot from it. It contains many additional resources that are extremely helpful. Thanks for creating it.
The main argument here is that at its current state, you have to invest not-so-small capital in order to be part of the bitcoin production system. If no one puts capital, no bitcoin will be created. In this regard, bitcoin is different from all other cryptocurrencies, including Ethereum (Yes, Ethereum also uses POW but there new blocks can be created much faster with less capital intensity). This has nothing to do with 21 million cap or scarcity.
Corrected. Thank you for pointing them out.
"what would really be useful is the valuation of a cryptocurrency relative to fields outside of it." Yes, this is something I want to explore more as part of the value of cryptocurrency in general. I am also aware of these types of projects, at least in the mission statement. But I really don't know if that is really something they want to do, or it is purely for the sake of attracting more investors.
I feel like the question of "what is the value of a cryptocurrency" should be different from "what is the value of cryptocurrency in general?". and I suspect that the answer will extend to beyond cryptocurrency.
For me, it is the latter question that I am interested in. I agree that if the question is about price, it is relatively straightforward. But I do believe that price is an important component of the valuation and the article you shared seems a good start for this angle. Thank you for sharing it.
Thank you for your comments. Those are all really good points and they inspired me with several new points that I need to think about:
- Value of something has objective (but not constant) components and high dimensional.
- A new questions is: Can there be a universal/fundamental component for the value notion such that we can use an equation like this: value of something = universal component + objective component to represent value notion?
- Regarding your last point, I can't help but think that when different people have different valuations for the same thing because of the human irrationality, what does that really imply? Does it imply there is a universal value notion for (some) things and people are bad at predicting that or there is no universal notion and people should have different valuations?
For instance, a computer's CPU is measured in GHz, which is a proxy for the number of calculations the CPU can run per second. So it is about one billion () calculations per second. Now let's suppose the number of calculations your program needs to run is , then you can make a Fermi estimation about the program's run time as , which is millisecond. Usually we would expect the actual run time will be within an order of magnitude of this estimation.
For Bayes Probability, Bayes rule is a great introduction. Yudkowsky’s intro also endorses this one.
I can relate myself to this post very much. I have not used a cell/smart phone for the past two years now, and I don't feel like I missed anything important in these years. I do have an office phone that I can use to call my family members several times a day. I faced the same problem in terms of setting up my university email earlier this year, but fortunately I could do it using office phone.
Every time when I had a problem like this, I tried to find a solution and usually I was able to find one. So I don't see myself getting back to a cell/smart phone anytime in the near future.
Great points. Two comments:
- Ability to generalize is something we would expect from deliberate optimizers. The point is that the strategies of accidental optimizers do not generalize well compared to deliberate optimizers.
- Deliberate optimizers intend to optimize and they do in fact optimize by applying one or more strategies. We would expect those strategies work well when a different agent applies those strategies to optimize the same criterion. In that sense, if the observer is this different agent, then yes.
If you know how the selection incentive works for a particular situation, you can exploit it to your benefit, or at least prevent yourself from being in an unfavorable position.
First example explains why sales people have such a bad reputation. For the second example, apparently products can't actually be good or bad by themselves. So it's the producer who makes the product. Are we assuming a situation where producer is able to select one time users from repeat users?
Can you elaborate your first example more? How does selection incentive come into play in those situations?
The second one is a good example for selection incentive: the incentives are there regardless of what we want. I like the counter intuitive actions in the third example: organisms are deliberately trying to achieve high reward by taking counterintuitive actions.
I particularly like the first & third one. John and I talked about incentive mechanisms in biological world (organisms, biological evolution, etc.) and these are really good examples of it.
Yes, "amplitude" is a more appropriate word. I've changed it. Thank you.
I disagree. The key factor to observe bullwhip effect is no coordination among players. For example, retailer does not share consumer demand information with the wholesaler, and the wholesaler does not share demand information with the distributor, etc.
Yes. The beer game has been frequently used in undergraduate and MBA programs when professors introduce the bullwhip effect.
It can be any real-valued measurement of objects, as long as we can reasonably assume the three assumptions are satisfied.
Thank you for your well-thought comment. One of the desiderata used to derive the original product rule is to use real numbers to represent the degrees of plausibility. So, it will be very interesting to see if the result still holds if we relax it to be a complex numbers.
I've taken the test not long ago and I agree it is quite informative. One concern I always have with these kinds of tests is that we tend to select options (at least in some cases) that highlight who we want to be rather than who we really are. I would like to know how this kind of bias is reduced.