Post-tutorial stats

jasonrohrer · 2018-07-02 17:15:54

Here are some stats from before and after the tutorial.

These show people who first played the game on various Saturdays back in time.

The stats are in this order:

Number_who_played_at_least_1_hour  /  Number_who_played_at_least_3_hours  /  total    percent_at_least_1_hour  percent_at_least_3_hours

The total is the total number of people who played their very first game on that Saturday, and the other counts are how many of those people played at least 1 or 3 hours total since then (in the intervening days or weeks). So the people from long-ago Saturdays should have an advantage here, because they've had more time to clock more playtime.

Sat June 30   51/36/61   83%  59%
Sat June 23   35/21/42   83%  50%
Sat June 16   33/25/47   70%  53%
Sat June 9    44/34/54   81%  62%
Sat June 2    89/62/110  80%  56%
Sat May 26    57/47/71   80%  66%
Sat May 19    46/27/58   70%  46%

Note that June 30 is the first Saturday where people HAD to complete the tutorial before logging any time in the game. The tutorial time does not count, and isn't logged. So 61 people got through the tutorial on Saturday (out of 65 people who bought the game). That's a lot of tutorial success, so that's good. And out of the 61 who finished the tutorial, 51 played at least one more total hour, and 36 played at least three more total hours.

On previous Saturdays, people were just dropped right into the game with no tutorial, so they were all logging life time from their very first moments in the game.

YAHG · 2018-07-03 00:24:49

I can't tell what this means? Did it work?

jasonrohrer · 2018-07-03 01:12:23

I'm not sure!

After I ran the data, I saw that it was inconclusive, but I posted it anyway.

I think that running the same analysis next week will give more info, because it will put this past Saturday on par with more distant Saturdays time-wise (i.e., how many people who bought the game have put in more than three hours a week later).

YAHG · 2018-07-03 01:27:29

I thought you were gonna do the data like you had before. That one with the player half-life being only a week.

Some interesting data you might like looking at before and after change, hopefully your data collection is detailed enough to reconstruct old data.

Average Generations to Die out
Baby life expectancy vs Generation#

I am imagining it as a graph with before and after change. That or a bunch of different colored lines so you
can see how it changes with changes you make to the game. I don't know if you have desired outcomes as
far as how often people die/live etc.

One way ANOVA tests are not super hard to do or learn how to do. I figure if you can program a game like
this you can do sums of squares and basic stuff like that no prob.

Analiyzing the variance before and after you can see if the change made a significant change statistically.
Given the hundreds (thousands?) or even more lives per day it should not be hard to get good sample sizes.

jasonrohrer · 2018-07-03 01:41:00

Well, yeah, I can do that, but it does take some amount of work.

The data is there, it just needs to be processed.

I wish I could easily make the data public, so that other people could run analyses.

The problem, currently, is that the accounts are hooked to email addresses. Those are the userIDs that connect one life to another.

I'd need to replace those with hashes or something before I released the data.

I guess I could do that nightly, as a cron job, and just copy the hash-replaced data files to a web folder for people to access as-needed.

YAHG · 2018-07-03 01:52:38

If you changed the account names into numbers like 1,2,3,4,5,6,7,8 etc. before you hashed them then I think that would
make it impossible to break as there would be no math based link between the email and the end hash. I figure order that
the account was made would be good enough. If time-stamping account creation is ok, then we can still see if accounts
came from x period of time like you can.

I don't see why you can't set this up to be done automatically either?

Also, I thought you did these pretty graphs for us cause you liked to..

I just assume tutorial will make people less noob. At least I won't have to teach my Eve Mom to make a hatchet by poking
the ingredients in order like I am a pre-born.

sc0rp · 2018-07-03 05:13:16

jasonrohrer wrote:

I wish I could easily make the data public, so that other people could run analyses.

Having access to lineage data would be great. I considered scraping lineage web site, but time resolution is very poor there ("starved 2 days ago"). And there is no connection across lives.

jasonrohrer wrote:

The problem, currently, is that the accounts are hooked to email addresses. Those are the userIDs that connect one life to another.
I'd need to replace those with hashes or something before I released the data.

Hash of (email + secret per user salt). No way to connect even known emails to accounts. People still can be somewhat deanonimized (you can look up their other lives), if e.g. they linked to their lineage from forum.

jasonrohrer wrote:

I guess I could do that nightly, as a cron job, and just copy the hash-replaced data files to a web folder for people to access as-needed.

Would be interesting data to look at.

JonySky · 2018-07-03 07:50:34

As I understood (my English is very poor, sorry) here shows us the percentage of people who have played at least 1 or 3 hours in total since I take the tutorial ... but this is very variable ... the tutorial does not affect the total time of a game session (there are very variable factors that change the time that a person intends to play ... for example, family life, friends, etc ...) I believe that he has to get the data of the life time that the people who have played have had (a very useful example, if a month ago the average of a new person's game was able to survive 20 minutes, but this week a new person can live 50 minutes) this means that the tutorial is working very well.

InSpace · 2018-07-03 12:33:46

I tried the tutorial today, it was good tbh, make a few more tutorials for some subjects, like cooking / smithing / farming so people get the base idea

One Hour One Life Forums

#1 2018-07-02 17:15:54

Post-tutorial stats

#2 2018-07-03 00:24:49

Re: Post-tutorial stats

#3 2018-07-03 01:12:23

Re: Post-tutorial stats

#4 2018-07-03 01:27:29

Re: Post-tutorial stats

#5 2018-07-03 01:41:00

Re: Post-tutorial stats

#6 2018-07-03 01:52:38

Re: Post-tutorial stats

#7 2018-07-03 05:13:16

Re: Post-tutorial stats

#8 2018-07-03 07:50:34

Re: Post-tutorial stats

#9 2018-07-03 12:33:46

Re: Post-tutorial stats

Board footer