[BPIP] Bitcointalk Public Information Project [Back in Action]

[Timelord2067]

No not one place - thirteen positions...  I've shrunk the image so you can see the rest of the text which you've quoted.

DT1 "election" at the beginning of May could have caused some trust scores to change. If someone with a score >= 70 dropped below 70 it could have helped you move up those 11 positions between April 25 and May 13.

Yes, that's true, I hadn't thought about the DT elections - it's also occurred to me just now that your scores might have a decimal place portion of the score that is not shown, hence a dozen people can appear to have the same score, but are ranked differently together with my score improving as the trust feedback positives have occurred in the last 12 months.

Would that be right?

[1713570990]

1713570990

[1713570990]

1713570990

[1713570990]

1713570990

[1713570990]

1713570990

[1713570990]

1713570990

[guigui371]

There are many instances of profiles having the same rank on the various lists.  If I can make a suggestion for the group to consider - say three people hold the rank of 21 e.g. instead of giving the next user the rank of 22, the next user is actually rank of 24:

Name      Score      Rank

Alice      1234      20
Bob      1202      21
Hank      1202      21
Sue      1202      21
Fred      1138      24

I think various sports such as the Olympics use that style of ranking.

Thanks for reading.

To jump back on this, I second that it could be useful to see this type of ranking implemented ( I know that the TODO list is huge).
Especially on the DT trust strong list
https://bpip.org/report.aspx?r=dtstrength
It appears that the last two members are  ranked 27.
However,  there are 91 people with a better trust "value" than them. 

Number of DT1 = 93
The last two (teeGUMES and Rmcdermott927) should be ranked 92 ex-quo

Nonetheless,
BPIP is a great tool.

[Timelord2067]

Nonetheless,
BPIP is a great tool.

Agreed.

Thanks for your support and feedback on this topic, I made mention a few posts back that there may be a decimal place that isn't being shown - I also wondered if perhaps the score/rank is being rounded up/down ??  It would go a long way to clarifying the ranks being equal but not equal.

[Vod]

You guys are talking about :
https://blog.jooq.org/2013/10/09/sql-trick-row_number-is-to-select-what-dense_rank-is-to-select-distinct/

RANK will duplicate numbers (if two people have the same score, they have the same rank).  I don't like that, because a secondary sort would rank one higher than the other.

DENSE_RANK is more realistic.  You cannot have two people in the tenth spot in a line up.   :/

[OmegaStarScream]

If you don't mind me asking (technical questions)... how do you manage to add newly created accounts to your database?

Do you just scrape the latest member from this page^[1]? If so, there is a possibility of missing a few accounts, isn't there?

Also, with this[2] being disabled and the search function returning 34 pages max, how can one scrape all the accounts? Do you just make a loop until you hit "The user whose profile you are trying to view does not exist"?

I might be able to make a list of banned users that will include bans before 2017. Maybe you and LoyceV will find it useful.

[1] https://bitcointalk.org/index.php?action=stats
[2] https://bitcointalk.org/index.php?action=mlist

[LoyceV]

If you don't mind me asking (technical questions)... how do you manage to add newly created accounts to your database?

For my complete List of all Bitcointalk usernames, I just download each profile and keep increasing the userID until the page doesn't exist.

[guigui371]

You guys are talking about :
https://blog.jooq.org/2013/10/09/sql-trick-row_number-is-to-select-what-dense_rank-is-to-select-distinct/

RANK will duplicate numbers (if two people have the same score, they have the same rank).  I don't like that, because a secondary sort would rank one higher than the other.

DENSE_RANK is more realistic.  You cannot have two people in the tenth spot in a line up.   :/

Well, this is were there a difference between front users and back users of a software / website.
I wouldn't know how to code it, changing between rank and dense rank.
I wouldn't be able to say which one is the best for all the existing features and future ones.

But, as a simple user, when there is a list of 100 people and that they are "ranked" using one variable, I would expect the last one to be ranked #100 or #99 ex-quo.

In the case of DT  ranking (my previous message), the current last one is ranked #27 out of a 93 pool of user ID.

[Timelord2067]

You guys are talking about :
https://blog.jooq.org/2013/10/09/sql-trick-row_number-is-to-select-what-dense_rank-is-to-select-distinct/

RANK will duplicate numbers (if two people have the same score, they have the same rank).  I don't like that, because a secondary sort would rank one higher than the other.

DENSE_RANK is more realistic.  You cannot have two people in the tenth spot in a line up.   :/

That's fine, I get that - My questions have been around how is it decided who amongst an equal score is given a ranking order.  My previous example where over 200 UID's have the same score but 200 successive ranks which then affect their most recognised scores.

[LoyceV]

I noticed this:

DefaultTrust has DT2 Strength: 41.
Where do you get the data? DefaultTrust is not shown on dtview and excluded from theymos' weekly trust data dump.
"DefaultTrust" occurs only once on the DefaultTrust Change Log.

[TheBeardedBaby]

How about parsing every account which wakes up? I think I suggested this before. We can avoid having cases like this > https://bitcointalk.org/index.php?action=profile;u=2655813
https://bpip.org/profile.aspx?id=2655813
If you are alredy parsing the seclog.php then the implemenation should be very easy.

[suchmoon]

Where do you get the data? DefaultTrust is not shown on dtview and excluded from theymos' weekly trust data dump.

It shows up here: https://bitcointalk.org/index.php?action=trust;full;dt

That's why it's shown as "DT2" on BPIP.

How about parsing every account which wakes up? I think I suggested this before. We can avoid having cases like this > https://bitcointalk.org/index.php?action=profile;u=2655813
https://bpip.org/profile.aspx?id=2655813
If you are alredy parsing the seclog.php then the implemenation should be very easy.

I think that's what we already do but I'll double check. The example you posted is something we need to clean up.

[TheBeardedBaby]

I think that's what we already do but I'll double check. The example you posted is something we need to clean up.

What you can do is to go to the seclog.php and check some of those who woke up resently. Some of them are parsed some are not. I don't know if the accoun have to post in order to be parsed or just have to wakes up. I didn't check that.

[TwitchySeal]

Not a big deal but I have a suggestion.

In your reports that have rankings, when there is a tie you are only incrementing the user(s) ranked one lower than the tie by 1.

I think they should be ranked by their actual position.

For example this:



Would become this:

[Timelord2067]

...

I'd certainly support that suggestion.  I'm sure others would too.

Kudos TwitchySeal.


Just a query, some recent occurrences should have caused an entry or two on this page for my UID, but they don't seem to be there.  Any suggestions what's caused that to occur?

[suchmoon]

It feels as if the excluded participant starts to clean his trust lists and get everyone out of there completely, but this is not the case. After a while, it all disappears from the log.

Is this a bug or a new BPIP feature?

It's a bug, i.e. willi974 did not actually clean out their trust list. It's caused by the fact that DT1 and DT2 lists on Bitcointalk are not always in sync. When someone drops out of DT1 they get crossed out on this list first:

https://bitcointalk.org/index.php?action=trust;dtview

And a few minutes later get removed from here:

https://bitcointalk.org/index.php?action=trust;dt;full

At which point BPIP doesn't know anymore of the two events are related and interprets this as if the person wiped out their trust list. You can see the difference in timestamps between when willi9974 dropped out and when their trust list was (incorrectly) shown as removed.

[TwitchySeal]

...

I'd certainly support that suggestion.  I'm sure others would too.

Kudos TwitchySeal.


Just a query, some recent occurrences should have caused an entry or two on this page for my UID, but they don't seem to be there.  Any suggestions what's caused that to occur?

Ha. Great minds think alike.

[suchmoon]

Just a query, some recent occurrences should have caused an entry or two on this page for my UID, but they don't seem to be there.  Any suggestions what's caused that to occur?

What entries did you expect that are missing? If you included someone at the DT2 level it may not show up. Basically anything that's not here: https://bitcointalk.org/index.php?action=trust;dt;full won't be on BPIP either.

[Timelord2067]

What entries did you expect that are missing? If you included someone at the DT2 level it may not show up. Basically anything that's not here: https://bitcointalk.org/index.php?action=trust;dt;full won't be on BPIP either.

I can see why you are puzzled, you've already skipped past my question when you answered here:

...

... after having removed the other entries.  

[sandy-is-fine]

Interesting bug:  Quadruplicates  None can be accessed.

[suchmoon]

Interesting bug:  Quadruplicates  None can be accessed.

I have removed the duplicates so it should work now:

https://bpip.org/profile.aspx?id=2817355

I'm still not entirely certain how that happens. There are a few other duplicates so I'll look into those for clues.