Basic research focuses on improving the understanding of a law of nature. Our company does basic research besides profit making just like many other Silicon Valley businesses. We disclose some results here. Hopefully these will be useful to others.

greenleaf

Copyright© Miklos Szegedi, 2022-2023.

Problem. How big is the government, and how much taxes should we pay.

Solution. The size of the government is the percentage of GDP that is spent on taxes. Interestingly this value strictly correlates to the essential services that are given to citizens. Essential services are provided even if the economy is contracting, so their size must be proportional to the total GDP whether it shrinks or grows. Essential services can be fulfilled by the government. It does not necessarily do so. It may rely on many grocery chains for example to do the supply chain, and help them out with taxes or interest subsidies from the taxes of other businesses, when needed. This allows the government to subsidize other causes like basic research from the rest of tax income. However, any non-essential government services are cut in favor of essential ones when tax revenues drop. The government may raise taxes to provide non essential services like green infrastructure. Businesses will compete and apply for these subsidies making costs lower. This will eventually reduce the spending from taxes or increase the GDP, so that the tax impact of such green investment becomes negligible. Essential services can be defined as a sustained and stable set of services that grow and shrink together with GDP in a long time. Societies can add or remove from the set gradually as they see fit. Any service that is stopped on GDP drop or that shows up on GDP growth temporarily are not essential by design. The purpose of tax collecting governments is to provide these essential services as a result. It is not redistribution, it is just responsibility ensuring that the basic services are working.

Problem. Machinery supporting work is priced compared to the labor it is attached to.

Solution. Imagine cloud services. An engineer may double their efficiency by buying cloud services that generate software, and run tests to improve reliability. They may hire more engineers instead. If the marginal cost of the cloud rental to software features shipped is less than the marginal cost of hourly labor, companies will invest more in these. This is what we saw with the rising revenue of cloud services like AWS or Azure in the 2010s. When labor is marginally cheaper, companies will hire more, who will spend time leveraging the existing capital, or use pen and pencil to design buildings. Example: if $1000/phone app of labor writes code with $1000/phone app of cloud services, we are in equilibrium. If cloud is just $900, companies will invest more in cloud, so that the labor use is more efficient. Such investment may not be renting or purchasing. It may be capital investment in differently skilled engineers to make compute better, more efficient, faster. B2B customers may rather invest in more employees than buying expensive equipment like a Cadence simulator to design semiconductor chips, if the marginal return in chip designs shipped requires it.

Problem. The value of secrets is zero.

Solution. The value of information that is not disclosed to all participants in a game, is a bond that is paid, when the information is disclosed. The opportunity cost of withholding information means new products will spread slower. The opportunity cost is that there will be a distrust of expectations of other information witheld.This results in less efficient markets that cause inequality, inefficiencies, risks and the resulting inflation and higher interest rates. Withholding information sometimes generates more income for one participant. Companies like to do trade shows as a surprise promoting new products. However, this gain is due to the show that it generates and the resulting press coverage that is higher than the distrust and loss of opportunity. Therefore, investors will favor such a company neglecting infrastructure investment elsewhere reducing the efficiency of the entire system.

Problem. Information that management needs for the best decisions is finite.

Solution. Information supports rational independent decision-making. The more information, the better decisions are made as a result. Information withholding does not have value as a result, companies will pay to get it by buying trade magazines, stock exchange information, and statistical data. Once all information is collected, a decision is made using a portion, maybe 20% using the Pareto theory. This means that 80% of information was not needed. Companies will pay sources like Gartner to filter the information keeping the important data. Confidential information about competition is rarely needed. Companies can easily assess the demand and supply in a market by posting a bid and checking the reception publicly. Public markets replace hoarding of confidential information as a result.

Problem. The Law Of Energy dictates that the variance of noise in the data specifies whether we need to collect more data and make consistent analysis of it.

The Law of Energy states that the quadratic value of measurements is stable across time and systems or dimensions. The sum of the quadratic values in two independent systems will always be the same as the sum of the quadratic measurements in the entire system. The Rule of Pythagoras proves this for two systems. The Rule of Fermat proves that there are no more independent systems than two that keep the Law Of Energy. A side effect of this in Physics is that teleportation may technically be possible between two independent systems only if the Law of Energy applies to the combined system, and there is only one way teleportation is possible between two independent systems and not multiple of them. If you want to teleport, you can find it only one way. Think about a helicopter rotor changing the altitude of the vehicle. Let's revisit this in case of information. If the mean expected return of a business plan has high variance, then we need to collect more information to explain the variance consistently from another system. This system can be composed of many systems but the Law of Energy, the variance of the sum of these variances equals the total variance and risk we measure. Think about a three-dimensional cube. We know we have to collect more data from the system with the highest variance. If we collect more data, we explain the noise, reducing the variance we see by combining the original and the newly understood system leaving some variance left of remaining systems. This is what we do, when we collect more Big Data, and this is what Machine Learning and Artificial Intelligence does to reduce the variance of the noise (the loss) by making a more consistent model understanding independent systems one by one at each epoch.

This article was revised on September 1, 2023.