Originally shared by Paul Allen
Google+ To Pass 10,000,000 Users Tomorrow (on 7/12)
As I promised on Saturday night, I have finished updating my Google+ membership model with new data and re-estimated the Google+ user base.
My surname-based analysis shows that the number of Google+ users worldwide reached 7.3 million yesterday (July 10) – up from 1.7 million users on July 4th. That is a 350% increase in six days. The userbase is growing so quickly that it is challenging for me to keep up, since the number of users of any given surname (even the rare ones I am tracking) seems to be climbing every day.
More impressive than last week's growth is the astonishing growth in users from yesterday at mid-day to tonight -- a 30% jump. My latest estimate tonight shows approximately 9.5 million users. This suggests that 2.2 million people have joined Google+ in the past 32-34 hours.
I project that Google will easily pass 10 million users tomorrow and could reach 20 million user by this coming weekend if they keep the Invite Button available. As one G+ user put it, it is easy to underestimate the power of exponential growth.
My model is simple. I start with US Census Bureau data about surname popularity in the U.S., and compare it to the number of Google+ users with each surname. I split the U.S. users from the non-U.S. users. By using a sample of 100-200 surnames, I am able to accurately estimate the total percentage of the U.S. population that has signed up for Google+. Then I use that number and a calculated ratio of U.S. to non-U.S. users to generate my worldwide estimates. My ratio is 1 US user for every 2.12 non-U.S. users. That ratio was calculated on July 4th through a laborious effort, and I haven't updated it since. That is definitely a weakness in my model that I hope to address soon. The ratio will likely change over time.
Since I have been tracking this same cohort of surnames from my first day, I am able to accurately measure growth over time.
I am not claiming perfect accuracy, but I do think the model is sound. A quant has suggested a mathematical formula that I can use to calculate a range of Google users with a 99% level of accuracy, and one of my employees is working on that math now. I hope to include that in future models.
Here is one way to look at my model. Imagine the U.S. government in 2020 has no money left. I know that's hard to imagine, but stay with me. Imagine they wanted to conduct a 2020 census and subsequent decennial censuses with a degree of accuracy (let's say 95%) and to do it on a shoestring budget.
They had complete data for 2010 - the population and growth rates for every city and town in the country. To do 2020, they could just take a random sampling of 100 cities and towns across the U.S. that were representative and conduct the census JUST for those cities every 10 years. If those 100 cities averaged the same growth rates as the rest of the country, then their decennial censuses would be fairly accurate but very inexpensive. (Obviously the US example won't work and shouldn't be tried, since the purpose of the U.S. census is in part to determine Congressional representation - so a complete census must be done in the entire country.)
But my project is like that - a low-budget sampling. I have randomly selected 100 uncommon U.S. surnames and I am tracking the number of Google+ users with those names - updating my counts every 2-3 days. I am assuming that the growth in G+ users with those surnames is similar to the growth in G+ users with the other 150,000 or so surnames in the U.S. If I had resources to include 500 or 1,000 surnames in my sample, then I believe my model would be more accurate. But my time and budget available for this project are small, so it is what it is. And then I take the 2.12 - 1 non-US to US ratio to complete the calculations.
I'm not sure how many more times I'll update this. I do believe it is quite accurate. Much more accurate than a guess. It is based on sound starting data, but some of my assumptions may not be perfect. I look forward to Google announcing actual user numbers, so I can stop working on this in all my spare time. Or, perhaps, someone will discover an advanced query that actually works - that returns unique user profile pages but no pages that contain posts. People keep suggesting queries will work, but so far, I have found that none of them is accurate for user counts.
For reference, here are my earlier posts on this subject:
4.5 million estimate on 7/9 (actually 12:15ish on 7/10)
https://plus.google.com/117388252776312694644/posts/1k85ZNPCu1A
1.7 million estimate on 7/4
https://plus.google.com/117388252776312694644/posts/VuKTMZm9xWy
https://plus.google.com/117388252776312694644/posts/1k85ZNPCu1A
Google+ To Pass 10,000,000 Users Tomorrow (on 7/12)
As I promised on Saturday night, I have finished updating my Google+ membership model with new data and re-estimated the Google+ user base.
My surname-based analysis shows that the number of Google+ users worldwide reached 7.3 million yesterday (July 10) – up from 1.7 million users on July 4th. That is a 350% increase in six days. The userbase is growing so quickly that it is challenging for me to keep up, since the number of users of any given surname (even the rare ones I am tracking) seems to be climbing every day.
More impressive than last week's growth is the astonishing growth in users from yesterday at mid-day to tonight -- a 30% jump. My latest estimate tonight shows approximately 9.5 million users. This suggests that 2.2 million people have joined Google+ in the past 32-34 hours.
I project that Google will easily pass 10 million users tomorrow and could reach 20 million user by this coming weekend if they keep the Invite Button available. As one G+ user put it, it is easy to underestimate the power of exponential growth.
My model is simple. I start with US Census Bureau data about surname popularity in the U.S., and compare it to the number of Google+ users with each surname. I split the U.S. users from the non-U.S. users. By using a sample of 100-200 surnames, I am able to accurately estimate the total percentage of the U.S. population that has signed up for Google+. Then I use that number and a calculated ratio of U.S. to non-U.S. users to generate my worldwide estimates. My ratio is 1 US user for every 2.12 non-U.S. users. That ratio was calculated on July 4th through a laborious effort, and I haven't updated it since. That is definitely a weakness in my model that I hope to address soon. The ratio will likely change over time.
Since I have been tracking this same cohort of surnames from my first day, I am able to accurately measure growth over time.
I am not claiming perfect accuracy, but I do think the model is sound. A quant has suggested a mathematical formula that I can use to calculate a range of Google users with a 99% level of accuracy, and one of my employees is working on that math now. I hope to include that in future models.
Here is one way to look at my model. Imagine the U.S. government in 2020 has no money left. I know that's hard to imagine, but stay with me. Imagine they wanted to conduct a 2020 census and subsequent decennial censuses with a degree of accuracy (let's say 95%) and to do it on a shoestring budget.
They had complete data for 2010 - the population and growth rates for every city and town in the country. To do 2020, they could just take a random sampling of 100 cities and towns across the U.S. that were representative and conduct the census JUST for those cities every 10 years. If those 100 cities averaged the same growth rates as the rest of the country, then their decennial censuses would be fairly accurate but very inexpensive. (Obviously the US example won't work and shouldn't be tried, since the purpose of the U.S. census is in part to determine Congressional representation - so a complete census must be done in the entire country.)
But my project is like that - a low-budget sampling. I have randomly selected 100 uncommon U.S. surnames and I am tracking the number of Google+ users with those names - updating my counts every 2-3 days. I am assuming that the growth in G+ users with those surnames is similar to the growth in G+ users with the other 150,000 or so surnames in the U.S. If I had resources to include 500 or 1,000 surnames in my sample, then I believe my model would be more accurate. But my time and budget available for this project are small, so it is what it is. And then I take the 2.12 - 1 non-US to US ratio to complete the calculations.
I'm not sure how many more times I'll update this. I do believe it is quite accurate. Much more accurate than a guess. It is based on sound starting data, but some of my assumptions may not be perfect. I look forward to Google announcing actual user numbers, so I can stop working on this in all my spare time. Or, perhaps, someone will discover an advanced query that actually works - that returns unique user profile pages but no pages that contain posts. People keep suggesting queries will work, but so far, I have found that none of them is accurate for user counts.
For reference, here are my earlier posts on this subject:
4.5 million estimate on 7/9 (actually 12:15ish on 7/10)
https://plus.google.com/117388252776312694644/posts/1k85ZNPCu1A
1.7 million estimate on 7/4
https://plus.google.com/117388252776312694644/posts/VuKTMZm9xWy
https://plus.google.com/117388252776312694644/posts/1k85ZNPCu1A
Megjegyzések
Megjegyzés küldése