Been thinking about ways to increase the accuracy, here are some extra thoughts and limitations. Feel free to add any too if you see a different perspective.
Generally, there are three macro perspective trends. Trend 1): Qubes is over-represented in a region or country. Trend 2): Qubes is at average represented in a region or country. Trend 2): Qubes is under-represented in a region or a country. If any region or data, falls into the trend 1, or trend 3, then it messes up the accuracy. Trend 1) speculated factors - Different culture (Can have huge influence). - Reasonable stable and functioning economy, towards a strong economy. - Peace. - Order and predictability in short term daily life. - Reasonable infrastructure towards great infrastructure. - Anything else that you can imagine in this, etc. Trend 3) speculated factors - Different culture (Can have huge influence). - Poor economy, country is not functioning well, or barely at all. - War with another country. - Civil war. - Turmoil and unstable government. - Poor infrastructure (roads, internet, food supply, reliability in expectancy). - Anything else that you can imagine in this, etc. Trend 2) is what we can calculate with pretty high accuracy given how physics work. However the real world is far more complex, trend 2) is not taking the many factors of life into consideration. The trend 1) and trend 3), as on the list above, have big influence. Similar problems are found in GNP (Gross National Product), which is something used by macro economists and politicians too, to measure how well a country is performing in its production. The drawback, just like trend 1), and trend 3) above, is the vast different cultures, history, current state, different ways from country to country on how to calculate, or even different ways in gathering the raw data used in the calculations, etc. The solution, is to limit these comparisons to the countrys own GNP from the year before, and to avoid comparing with other countries, unless, of course, the country look a lot alike in the trend 1) and trend 3) factor lists. For example USA states, may draw better similarities between similar looking states, compared to if you compare a US States GNP with say, Germany, Russia, China, Italy, and so on, whom have similar, but yet also very different cultures and factors that make comparisons inaccurate. The solution therefore, is to only compare where it makes sense to compare, either by comparing to your own GNP the year before, or only compare with a country that looks a lot alike. Keeping in mind that even within USA, a US state can be very different from another US State, so one has to be very careful with comparisons like these. Even if comparing a countrys own GNP from several years back, ones own country culture will likely have changed, and even the method of calculation, or method of data collection, can be different if going too many years back in the same country. However, if you do like inflation calculations, you can go year by year, one at a time, make % comparison with the countries own GNP, only one year back at a time. This way, you can see a chain reaction, only looking at small changes at a time. But its dangerous to try jump too far in the timeline, unless changes in trend 1) or trend 3) are taken into account. Given the complexity, this is notoriously difficult to do, in any way that represent accuracy. Even getting a close estimation can easily be notorious. So the takeaway? Reducing complexity, and limit ourselves into how we use and take the data for granted. For example, be mindful of all the various ways the data can be shaped differently from what reality really looks like. So keeping these challenges in mind from economics, we can draw a bit from it in our Qubes demographics. For example, if you know how many Qubes users are in the USA, or in China, EU, Africa, Russia, or any other similar region, which is very different to the rest of the world, yet similar inwards towards itself and its own culture, then we can increase the accuracy quite a bit. The problem is we don't have such data, and it probably isn't a good idea if the Qubes team start to look into the unique IP's in an invasive way. It's already troubling enough that they keep logs of everyone's IP to begin with. So what else can we do? We might be able to incorporate some secondary data, i.e. find out how many people live in a country without infrastructure. Then we can take the world population, and subtract the amount of people whom have no or extremely poor infrastructure. Another method, which can be used in addition to the above, or any other similar subtractions, is to figure out how many children and teenagers, as well as old people, there are in the world. While some old people, and likely some teenagers too, use Qubes, the bigger population of Qubes users are probably in the years of maybe, say, 20-50 years of age. It's a bit inaccurate to guess like this, but its even more inaccurate to include the age groups that likely don't use Qubes. So while not entirely accurate, at least, we can move closer towards accuracy. What else? Maybe we can find data on how many Chinese use Linux, for example, and then deduce by that, whether Chinese may be likely to find Qubes interesting or not. If we can find such reliable data anywhere. Still not too accurate to deduce in this manner, but it's way more reliable and accurate, than just making wild random guesses out of personal opinion. Once more, pushing closer towards accuracy. The more we can do of these accuracy measures, the more close we can get to the real life like numbers. Especially if we can find precise numbers on the big factors, like the ones with the number of people living in poor infrastructures, or the number of people of different age groups to filter out. Lets imagine, we came to the conclusion that at least 4 billion people (4000 million), can use Qubes, the remaining 3.7 billion (3700 million) people cannot. Okay, so here we go: - Analysis - Step 1,A) 82M German pop. divided by 4.000M world pop. = 0,0205 (or 2,05 %). Step 1,B) 1,330M Munich pop. divided by 4.000 world pop. = 0,000332 (or 0,0332 %). ---- Step 2,A) 25.000 Qubes users multiplied by German/world population ratio 0,0205 = 512,5 German Qubes users. Step 2,B) 25.000 Qubes users multiplied by Munich/world population ratio 0,000332 = 8,4 Munich Qubes users. So the more accurate we try to filter away any population that cannot use Qubes, the more we are left with a population with the potential to use Qubes. It does not matter if everyone uses Qubes in this population, what matters is that they got the potential to use Qubes, so that the differences are taken into account. By removing age groups unlikely to use Qubes, and removing poor infrastructure populations, then we can already increase the accuracy by quite a bit. The question is, how far can we go, and still be getting closer towards accuracy? For example, we can be trying to trim too far, and end up on the other side of the extreme. Say we deem a country to have a poor infrastructure, yet it still happens to have a significant amount of Qubes users due to privacy concerns from their governments. Or alternatively, children/teenagers who are smart, and venture into using Qubes, or even older people keeping up with times. It's not so simple, but, I think we can at least shave off a few billion people. The question is much more, when is shaving off, too many? Is 3billion too much? or too little? etc. -- You received this message because you are subscribed to the Google Groups "qubes-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/qubes-users/c7b83fc1-853f-469e-8483-9453101b8039%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
