1. Introduction.It’s a rather dull and rainy day where I live, so contrary to the household spirit of what can be done on such a day, I’ve grabbed my laptop to take a look into something I was curious about. In the light of the forum encouragement for us to build our Custom Trust Network, I’m keeping track of how these are evolving.
So far, we’ve got:
<…>
Date Trust/Dist. rel. Trust rel. Distrust rel. Customized Lists Distinct Trusted/Untrusted[b]
19/01/2019 25387 21811 3576 4052 8662
12/01/2019 23800 20720 3080 3957 8498
05/01/2019 22605 19637 2968 3891 8527
That is an additional 95 customized lists this week (rather a small amount), 1091 Trust relations and 596 net Distrust relations.
The breakdown of the existing Custom Trust List by rank is as follows:
Truster:
nfromRank nUsers %
VIP 8 0,20%
Donator 16 0,39%
Administrator 2 0,05%
Global Moderator 2 0,05%
Staff 16 0,39%
Legendary 514 12,69%
Hero Member 483 11,92%
Sr. Member 665 16,41%
Full Member 664 16,39%
Member 549 13,55%
Jr. Member 84 2,07%
Newbie 1016 25,07%
Copper Member 25 0,62%
(non derived) 8 0,20%
Total 4052 100,00%
The profiles are derived from my weekly merited scraped profiles and @Piggy’s latest full database dumps (early December 2018). Therefore, some profiles may have ranked-up in the meantime. The “non-derived” are some I need to look into (they do not match when crossing data due to character issues in different codepages). The Copper Members are those that are derived from @Piggy’s dataset, and that I have not tried to derive their core rank (I do on the merited profiles).
The weird thing that stands out is the amount of Newbie accounts with a Customized Trust Network. I manually checked some of them and they are still Newbies. It does not mean that they are recently created Newbies, and we can see that somewhat further down.
Trusted/untrusted:
nToRank nUsers %
Founder 1 0,01%
VIP 17 0,20%
Donator 45 0,52%
Administrator 2 0,02%
Global Moderator 3 0,03%
Staff 23 0,27%
Legendary 1124 12,98%
Hero Member 1117 12,90%
Sr. Member 1435 16,57%
Full Member 1341 15,48%
Jr. Member 164 1,89%
Member 1026 11,84%
Newbie 2246 25,93%
Copper Member 105 1,21%
(non derived) 13 0,15%
Total 8662 100,00%
Here again, the Newbie segment stands out weirdly.
3. Custom Trust Lists by Date of last active.The profile date of last active could well be a good indicator as to whether the Custom Trust Lists are recent or old. Here I’m deriving last active in the same way as rank (so it could be off by a month at most in the worst case, being therefore not exact).
yearLastActive LastActiveQ nUsers
2019 Q1 529
2018 Q4 330
2018 Q3 1267
2018 Q2 434
2018 Q1 324
2017 Q4 100
2017 Q3 215
2017 Q2 126
2017 Q1 67
2016 Q4 17
2016 Q3 64
2016 Q2 56
2016 Q1 67
2015 Q4 25
2015 Q3 50
2015 Q2 110
2015 Q1 47
2014 Q4 16
2014 Q3 84
2014 Q2 63
2014 Q1 25
2013 Q4 7
2013 Q3 18
2013 Q2 3
(non derived) Q 8
Total 4052
Ok, this is interesting. Only a small fraction seems to have a recent (2019 Q1) last activity date on their profile. What I derive from that is than the vast majority of existing Custom Trust Lists have not been update recently (nor are the owners of the list actually online recently).
I’ve gone and checked the above shown Newbies with a Custom Trust List
yearLastActive LastActiveQ nUsers
2019 Q1 6
2018 Q4 65
2018 Q3 259
2018 Q2 115
2018 Q1 102
2017 Q4 21
2017 Q3 77
2017 Q2 44
2017 Q1 24
2016 Q4 4
2016 Q3 24
2016 Q2 18
2016 Q1 23
2015 Q4 19
2015 Q3 20
2015 Q2 58
2015 Q1 17
2014 Q4 11
2014 Q3 42
2014 Q2 36
2014 Q1 11
2013 Q4 7
2013 Q3 11
2013 Q2 2
Total 1016
Here I’ve tried to see how many forum members are actually included in a Custom Trust List (either trusted or untrusted) by groups:
nTrustSize nCases %
>10 553 13,65%
10 58 1,43%
9 93 2,30%
8 85 2,10%
7 101 2,49%
6 134 3,31%
5 144 3,55%
4 283 6,98%
3 404 9,97%
2 590 14,56%
1 1607 39,66%
Total 4052 100,00%
Ok, so 64,19% of Custom Trust Lists have between 1 and 3 trusted or untrusted members; rather small. Likely the trust scores on that those profiles will see are rather limited (the complete feedback will be there to see though).
We could go further on (do this table by number of Trust and Distrust relations, cross by rank, etc.), but with all the above I think we get a general idea how Custom Trust Networks look right now.
5. Additional stuff Additionally, we can play around with:
Link to the
Personal Trust Network.
Link to the
Base File (Data to load the Personal Trust Network (first tab) + Summary per user (second tab -> interesting to see most trusted/distrusted and so on)).
