Based on the slowing rate of membership growth, I decided to fit an exponential rise curve to see how well it fits, and what it might predict.
The curve that best fits the data is one that has the formula
y = 6024 * (1 - exp( -.1391 * x ) ) where x is months since introduction.
This is a curve that asymptotically approaches an upper limit of 6024, and reaches 6000 at roughly 36 months.
Plotting the actual values on the curve shows a relatively good fit.
I actively pursue new people to request Add Friend (rewarding me with 2 myASQ points) and new posts and files to Like and Download (rewarding me with 1 myASQ point).
Yesterday myASQ experienced unusual growth (over 100 new members). This was explained by Cynthia Nazario as having a new community automatically imported into myASQ. If more surges like this occur, then the enrolment of members in myASQ could reasonably exceed 6,000 members.
I also think that the current totals are reflective of the low percentage of activated geographic and technical communities within myASQ. Once a higher percentage are brought online, their membership will follow.
I also commend the dedicated ASQ staff members like Gretchen Peterson , Trish Borzon, and Seiche Sanders for their promotion of ASQ content and publications, which draws in positive posts from others and reflects the best characteristics of ASQ. In 2019, I will shift my focus from the controversial aspects of ASQ's administrative minutiae; redirecting and pivoting back to the excellent content, publications, and individual member benefits which ASQ provides with more abundance than any competing society or association for Quality professionals and practitioners.
This is a classic example of how two reasonable people can reach opposite conclusions from the same data. When I saw the comment about an automatic upload of multiple new members I interpreted it as artificial inflation of membership, since these new members are not voluntary participants. Thus I would conclude the true number of participants is being overestimated.
But it's just a model, and an early one at that.
As we learn the best ways to use and feature myASQ, you will see additional areas where we will incorporating 'conversations' in different areas of the web.
It's all very exciting! I love the catching up on what members are looking for, their needs, and wants. What a great tool to capture the Voice of the Customer!
I am pleased to introduce Harry Rowe as one of the resident experts on VoC within the ASQ context. If myASQ wants to be a VoC interface, they would do very well to engage Harry's considerable proven experience. You may already be aware of his work, so if I am preaching to the choir, I apologize :)
I think myASQ is still in the Storming and Norming phases, as there are still boundary issues being determined and clarified.
Please don't take the slide rule comment I made to be an ageist thing. I kick myself for never learning how to use one.
My only intent is to show my great admiration of you and your service to ASQ and the fact that you make your claims based on facts and data and actual statistical analysis which seems to be kind of a lost art even with Quality practitioners.
However... when the robots with their artificial intelligence finally reach the point of singularity and try to exterminate the human race by disabling our technologies, I will be depending on you and your trusty slide rule to save us, John Connor, oops, I mean Harry Rowe.
I even created a new tag on myASQ - "Harry will save us" ;-)
Funny. I still have some of those slide rules from the "olden days" (late 1960's). The bamboo ones from Post and K&E were considered the best, but the machined aluminum ones from Pickett were making inroads. (They didn't have to be adjusted when the humidity changed.) But I started programming computers in 1968. When I undertake a math or statistics problem now, I usually resort to R (free download for Mac, Windows, or Linux). Machine learning is one of my current interests.
One of my nicknames (used affectionately, I hope) is "older than dirt". No offense taken.
An exponential curve still fits the actual data quite well. The model that best fits the data now is
6398 * ( 1 - exp( -0.128 * months ) )
That is a curve that asymptotically approaches 6398 and reaches 6000 at approximately 23 months. This is slightly more optimistic than the maximum of 6024 with 6000 by month 36.
7017.4*( 1-exp(-.112*months) ).
That is a curve that asymptotically approaches 7017, first crossing 7000 in month 54.
Over the last three iterations with additional data points, the terminal value has increased slightly from 6017 to 6398 to 7017, while the exponential factor has decreased from .139 to .128 to .112.
If you're wondering why I haven't provided an R-squared value for the fit, it's because R-squared is not a good metric for nonlinear regression.