Mark Edward Campos

Fact-checking Mailbox

Infographic

No Comments


Share this post

Mailbox App has gained a lot of buzz in the last few days around the net.  Are they failing with their slow rollout?  They launched with a controversial and possibly brilliant queuing system to manage server loads.  This means hundreds of thousands of people have a glorified countdown app.  Hundreds of one-star reviews and angry tweets later, Mailbox official has been directing people to this graph on their website:

This graph screams fake math to me.   With so much hype, Mailbox, I assumed, would err on the side of more-information, right?  Like many others, I’ve become nearly religious in opening the app and watching the numbers fall.  I’m a glutton for numbers though.  I realized fairly quickly when comparing queue numbers with a friend that the queue isn’t faked in the app.  It’s the real deal – my position in line + the ‘people behind me’ added up to a brand new registration.  A) kudos to mailbox for releasing normally sensitive user registration numbers and B) I can verify this silly graph on the Mailbox blog and verify their supposedly nonlinear fill rate.

Over the last 3 days I’ve opened the app 34 times and taken 34 data points.  Data here.  I threw this into RStudio and this is what I got, plain and simple:

I was shocked to see that rate has remained 100% consistent during all of feb 8 and 9, and actually DECREASED late Feb 9th.  Today things seem to be back to the exact same rate as late last week.  Too bad, for a system with so much hype.  So when will I (reserved via text message weeks ago) see my Mailbox?  At this rate, next week sometime.  Here’s a graph showing linear-rollout in blue ( no reason so far to believe otherwise ) and user-registration numbers in red.

– Mark

 

PS. Analysis is very simple with lubridate and ggplot2. R code for full transparency here:

library(lubridate)
library(ggplot2)
mail <- read.csv(‘mailbox.csv’, header=FALSE, col.names=c(‘date’,’me’,’them’), sep=’,’)
mail$date <- ymd_hms(mail$date)
ggplot(mail, aes(x=date, y=me)) + geom_point()
ggplot(mail, aes(x=date)) + geom_point(aes(y=me, color=’red’)) + stat_smooth(aes(y=me), method=lm, fullrange=TRUE) + ylim(0,550000) + xlim(x,y) + geom_point(aes(y=them, color=’blue’))

0 Responses to this post
Add your comment