Deflation and "Selected Best Nodes"

[LanX]

Hello Brothers

when bored I start reading and enjoying the wisdom of "Selected Best Nodes of All Time, 50 of the top 2000 nodes"!

I don't know if I'm the only one noticing¹ but it's pretty difficult to find nodes written after 2006!

As I'm looking at it now the last node N° 50 is "DNA is Life" with 99 upvotes, which pretty much explains the phenomenon.

It's the deflation (the opposite of inflation) of votes in the last years. IAW votes were "cheaper" back then.

The best node of 2013 apparently was "perlmonksflair.com : get your perlmonks flair!" with 87 votes.

I certainly doubt that the loss of monks or votes in the last years represents a decline of quality.

My suggestion (IFF easily realized!) is to select the nodes after multiplying a antiproportional weight factor which is derived from either average votes or maximum votes in the corresponding year.

This is far from being urgent, I just thought it might be worth noting, that ATM "selected best nodes" only represents the far past of the community.

PS: another factor could be that old posts had more time collecting votes, I don't have the necessary statistical data to check. Anyway it's still strange if this only applies to nodes written prior to 2006.

updates

¹) older discussions to the same topic:

• Selected best nodes - nothing recent, why?

[eyepopslikeamosquito]

This old node shows reputation ranges for the top twenty best nodes of the year near the end of the year:

• 2007: 115-238 votes
• 2008: 105-180 votes
• 2009: 87-141 votes
• 2010: 72-144 votes
• 2011: 60-98 votes
• 2012: 58-77 votes
• 2013: 56-87 votes
• 2014: 55-86 votes
• 2015: 52-76 votes
• 2016: 49-66 votes
• 2017: 50-58 votes

[LanX]

Thanks a lot,I tried finding older discussions but strangely missed this one... I'll update the OP. :)

[tangent]

I think I've looked at almost every one of those old nodes by now and I've learned quite a lot from them, and not just about perl. These nodes could be left for new visitors but it would be great to have an option to see more recent "best nodes" as I'm sure there is a treasure trove of information in there.

[LanX]

I agree, enlarging the "population" wouldn't mean a loss! :)

maybe best 5000 or even 10000 to choose from?

[kcott]

G'day LanX,

I concur with what you're reporting and would also like to see the selection including more recent nodes.

Perhaps one of the simplest solutions would be to have additional lists that excluded older timeframes; for example, "A selection of 50 of the top 2000 nodes from the last 3, 6, 9 and 12 years". [Note: I chose 3, 6, 9 and 12 arbitrarily]

Another possibility might be to have a means to make a custom selection: "A selection of W of the top X nodes from years in the range Y to Z". I appreciate that the processing required to provide this feature may make it in impractical.

There might also be a more acceptable middle-ground between those two suggestions, e.g. "A selection of 50 of the top 2000 nodes from years in the range Y to Z" with specific custom selections being cached and retained for 24 hours.

Anyway, those are just ideas for monks to mull over. Beyond wishing to see a wider selection of "Selected Best Nodes", I'm not too fussed how that might be implemented.

[Limbic~Region]

LanX,
For nearly a decade, I was in a torrid love affair with Perl and the Monastery but have not been active for the past couple of years. Here is what I think is going on:

A new node has a relatively small window of time (newest nodes - recently active threads) to amass the majority of its votes. If it does not reach a certain threshold, it is lost to the archives forever. That threshold continually increases because older nodes that were above that threshold and are still visible through Selected Best Nodes slowly accrue new upvotes dragging the "norm" up. Here is why I think this is what's going on:

When Selected Best Nodes was first introduced, the 50th randomly selected node had a reputation somewhere near the low 70's. Over time, I watched that number climb. This morning I saw that it is nearly 100. While I believe many extremely experienced monks have moved on from PerlMonks - there are still many quality nodes written that would likely have been in the best nodes if they had been written years ago.

For what it's worth, my memory of history is that Selected Best Nodes was created to fix this exact behavior because prior to that only the very top nodes could be seen.

[LanX]

Hi Limbic~Region,

this effect surely exists, I myself remember that a younger (XP-whoring) me blindly up-voted such nodes one or two times. (I was a young monk and hungry for reputation ;)

But like proven by eyepopslikeamosquito there is certainly a vote-deflation effect involved.

Selecting nodes from all years best should improve the situation, at least to a point where age doesn't matter.

For completeness (this over-extends the topic):

Other possible selection criteria for other compilations could be :

• the number of links into a thread ( compare google ranking) or
• tags used (unfortunately is the tag-nodelet mostly ignored).

[oiskuu]

Seconded! Many of the old articles have not aged well, or have lost their relevance (e.g. ancient bug reports). Perhaps it would be sensible to allow nodes to "settle in the dust" somehow. Scale the probability of inclusion by some factor of hotness? Last upvote time?

[salva]

Or divide the total number of votes by the number of days since the entry was posted to get the average votes per day.

[shmem]

I don't know if I'm the only one noticing but it's pretty difficult to find nodes written after 2006!
As I'm looking at it now the last node N° 50 is "DNA is Life" with 99 upvotes, which pretty much explains the phenomenon.
It's the deflation (the opposite of inflation) of votes in the last years. IAW votes were "cheaper" back then.
The best node of 2013 apparently was "perlmonksflair.com : get your perlmonks flair!" with 87 votes.

We'll all muse about flairs when we grow old, and get to younger nodes once we upvoted all the brilliant nodes of those giants on whose shoulders we stood a full working lifetime. Pretty soon it will be difficult to find Best Nodes Of All Times written after 2021!

[Anonymous Monk]

I can't recall even a single node written after 2021 that was worth upvoting at all.

[Your Mother]

2021 was actually a banner year for all the 8 perls. You don't remember because you developed a pretty serious case of Alzheimer's in 2019.

[pemungkah]

Can I have a ride in your time machine?

And is Perl6 done?

[LanX]

> all the brilliant nodes of those giants

true, nothing better than several Perl OO discussions from 2002... :-}

(pity Your Mother changed her user image again, I'd love to link to it now ...)

[Your Mother]

I put it back for now since I don’t have a canonical external source for it that wouldn’t spoil my surprise.

[Skeeve]

I'm wondering how those nodes are selected. I have a node from 2005 which, according to the "Rep" be should at position 32-35 but it's not listed at all.

[LanX]

AFAIK it's a daily random selection of 50 from the top 2000.

So wait a month our so and check regularly.

(well if your claim is justified, I didn't want to vote down just to see the points given ;-)

update

Strangely the clear description is hidden in Best Nodes

Also check out Selected Best Nodes — today's random selection of 50 of the top 2000 nodes of all time.

[Discipulus]

and what about the simpler solution: best nodes by year?

L*

[LanX]

The fun of this category is to have a random selection, otherwise it gets boring and people stop checking. :)

But I think it's worth considering to simply add the top n (50? or 100?) nodes of every year into the pool, together with these all time bests.

Like that "era effects" are mitigated.

And it's easy to implement.

[McA]

Hi Rolf,

probably it's an indicator for the following:

Everything gets worse...

;-)

Regards
McA