Topic: Customized Trust Network – Interactive tool to see who we trust/distrust - page 4. (Read 2263 times)

It doesn't didn't but I was looking at some shady individuals so overestimating their chances was ok in the context.

Thanks DD for the update. Even awesomer now.

No … it does not exclude those merits sent from oneself (I has version 1.0 of @theymo’s OP in mind). If I skip my "siesta", I can take a look to see if I can exclude those merits (it does seem rather more useful if I manage to exclude them and adjust it to that criteria).

Note: If the trust.txt file is update more frequently I’ll probably stick to a weekly update just the same, in order to align it to the merit update frequency (and manage to get some free time …).

Edit: Done. nFromEarnedMerits should now reflect the earned merit by the FromUser, minus the earned merit received from the ToUser in the Trust relationship.

I'm amazed how many people have included me Thanks for making this.
I've asked theymos (here) if it's possible to update trust.txt more frequent. Considering the many changes people make at the moment, it would be nice to be able to see them faster.

I assume that doesn't exclude the Merits sent by the user himself, right? Because theymos excludes those:

Added a second tab to the Base File in order to include a summary per user:


Sort, filter and search as you like.

Note: I may throw Added some additional merit related columns to the summary tab, after seeing @suchmoon's post.

Awesome sauce. I can choose nFromEarnedMerits filter 10+ or 250+ to see potential votes.

Looking good:


Edited 2020-11-30 to fix a broken image

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))

This is the last update: This product will be discontinued on the 03/12/2019

Data as of 30/11/2019 (trust file is released on Saturdays and not Fridays as is the merit.txt file).

Summary:
note: skipped week -> 20/07/2019

Date->                                       Date of update.
Trust/Dist. rel. ->                      Number of trust/distrust relations in the file.
Trust rel. ->                               Number of Trust relations.
Distrust rel. ->                           Number of Distrust relations.
Customized Lists ->                   Number of customized Trust lists.
Distinct Trusted/Untrusted ->  Number of distint user profiles that are either trusted or untrusted by others.

Profiles that all trust each other:
-   7 members where all trust all: 5 cases (complete graph networks)
-   6 members where all trust all: 39 cases (complete graph networks)
-   5 members where all trust all: 179 cases (complete graph networks)
-   4 members where all trust all: 563 cases (complete graph networks)
-   3 members where all trust all: 1.118 cases (complete graph networks)
-   2 members where all trust all: 2.369 cases (complete graph networks)

In addition:
- Analysis – A quick glance at the current Custom Trust Networks.
- Analysis - Reciprocal trust, distrust and asymmetries + all trust all subgraphs.

---

1. Introduction.

Back in April 2018, I created a similar project to the one I’m going to describe here, but left it aside since, at the time, it didn’t seem to be useful enough to endeavour a weekly update of the information (see A picture of what our Trust/Distrust Network looks like). Nevertheless, I’ve decided to retake it in light of the recent changes, modifying and adding a few things that were not relevant in version 1 of the project.

The base that I’m using here is the weekly forum official trust file (https://bitcointalk.org/trust.txt.xz), which basically includes a set with the following information:
a->b    'a' trust 'b'
a-/>b   'a' does not trust 'b'

As far as I’m aware, the file includes only data relative to customized trust lists, so DT is not shown here.

What I have done is include certain information for both ends of the 'a' and 'b' trust relation, so that we can see for a given user:
- The set of forum members 'a' trusts and/or doesn’t trust.
- The set of forum members who trust 'a' and/or don’t trust him/her/it.
The included columns are: user names and user ids involved, urls, rank, trust type, earned merits.

A few things to consider:
- The tool is visual and filterable, but it is a Beta tool created by Google, that will be discontinued in December 2019.
- The tool does have some bugs, and many limitations (I can barely add anything more to it in terms of functionality).
- Once you get the hang of it, it works fine, but it takes a bit to get used to the filters and specially resetting them.
- The list that serves as a base for the "Data" tab could suffice really (but we lose out the graphic part if we do).
- Rank and IDs are derived from a combination of the data I download (merited profiles), but also from @piggy’s full dataset (for the data not related to merited profiles).
- A few (12 o so) profiles with weird characters are not ranked nor merit counted (I will fix them briefly - I hope).

Note This is a similar product in terms of usage to Our very own sMerit Network Picture.


2. Data.

The Data Tab contains all the available data I’ve managed to group up. Each row contains:


  FromTo: The Trust relationship as found on the trust.txt file.
  UserfromName:  The username that trusts/distrusts.
  UserToName: The username that is trusted/distrusted.
  TrustType: Y: trusts; N: Does not trust.
  nFromEarnedMerits: Earned merits by the user that trusts/distrusts (not including merit sent between them).
  nToEarnedMerits: Earned merits by the user that is trusted/distrusted.
  nFromdummyWeight: For internal purposes (equal to nFromEarnedMerits +1).
  nTodummyWeight: For internal purposes (equal to nToEarnedMerits +1).
  nFromId: The User Id that trusts/distrusts.
  nToId: The User Id that is trusted/distrusted.
  nFromRank: The rank of the user that trusts/distrusts.
  nToRank: The rank of the user that is trusted/distrusted.
  nFromURL: The Bitcointalk profile URL of the user that trusts/distrusts.
  nToRank: The Bitcointalk profile URL of the user that is trusted/distrusted.

All trust/distrust relations are shown (until I encounter a volume problem that may well happen at some point). You can browse through this dataset, and apply filters.



You can add multiple filters, and multiple values to them. If at some point you get lost, it is best to delete the filter with the cross mark and re-add the filter by selecting it from the blue Filter dropdown menu.


3. Who I Trust/Distrust.

On this tab you can see a graphical representation of your Trust Network, specifically of those you trust. Simply enter the username on the "UserFromName" filter, press "find" and the graph should show the trusted/untrusted network for the selected user.
Note that both trusted and untrusted are shown, but I cannot give the nodes a different color. In order to distinguish between the Trusted Network and the Untrusted Network you will need to additionally filter by "TrustType" (Y/N).

The size of the nodes are proportional to the number of earned merits a user has.


I have used @theymos's Trust network for my examples, since mine is rather limited so far. In this case, I've set the filter to show those trusted by @theymos (TrustType = 'Y').

Unsolvable bugs:
-   Not all nodes are shown by default necessarily. You will need to type a number of nodes (top left of graph) higher or equal to the max nodes shown, in order to see them all.

-   You cannot see all the underlying data here by pulling down the three button handle above the graph (rows are all shown, but only 5 columns, and I cannot see how to change that). To see the actual underlying data, you’ll need to change to the Data Tab and filter it there as explained before.

 
4. Who Trusts me/Distrusts me.

This tab works in the same manner as the previously described tab. The focus here is on the "UserToName", being that parameter the name of the user for whom we want to see the network of people that trust or distrust him. As before, we need to filter by "TrustType" to delimit the graph to those that either trust us or distrust us.


In this case, I've set the filter to show those untrusted by @theymos (TrustType = 'N').


5. A few interesting numbers.

-   Trust.txt files contains 23.800 trust/distrust relations (22.605 last week).
-   Trust relations: 20.720 (19.637 last week).
-   Distrust relations: 3.080 (2.968 last week).
-   People that have a customized Trust Network: 3.957 (3.891 last week).
-   People that are included in a customized Trust Network: 8.498 (8.527 last week, which is larger – likely due to list resets).

 The changes a really new, but in spite of the Meta hype, not that many new people seem to have created a customized list yet. Well see how those numbers evolve.