- This topic is empty.
Nov 3, 2006 at 7:26 am #1366114Eric NobleBPL Member
@ericnobleLocale: Colorado Rockies
I set my personal message email address last night and received a email confirming the fact but when I looked at my profile there was no indicator that it had been set or what it was set to. I deleted the confirming email before I realized this. At some point I will forget what I did. The email account I am using should be visible to me in my account or at least an indicator that it is set.Nov 3, 2006 at 12:04 pm #1366138Glenn RobertsMember
@garkjrLocale: Southwestern Ohio
How’d pj – or anyone else, for that matter – get a higher rating than Ryan? (pj: yours just happened to be the one I noticed, my friend; it was probably that hair-shirt you’re wearing!! You know I’m only kidding – it’s a great avatar.)Nov 3, 2006 at 12:34 pm #1366142
Glenn, Hair shirt? What hair shirt? That’s my chest hair!! Just washed it and can’t do a thing with it.
Oh, and about the numerical ranking, Ryan Jordan nailed it; see his post where he said, referring to another’s comment similar to yours about his and mine numerical rankings, that the “website is broke”!!! That explanation sure satisfies me.Nov 3, 2006 at 2:50 pm #1366153Siegmund BeimfohrBPL Member
Rank numbers are mysterious! Mine:
BPLRank (All Time): 2.49
BPLRank (This Year): 2.65
Only joined in late August (although have been reading articles/posts for many month) so not sure why two numbers should be different??Nov 3, 2006 at 3:16 pm #1366159larry savageSpectator
@pyeyoLocale: pacific northwest
It appears that the rank is a simple ratio of insightful responses to b.s. responses times the inverse of the amount of items purchased at the bpl online store.
This said, my bpl rank should now drop significantly.Nov 3, 2006 at 3:50 pm #1366165
>> Only joined in late August (although have been reading articles/posts for many month) so not sure why two numbers should be different??
That is an extremely insightful observation, Sig. Good work. ;)Nov 3, 2006 at 6:55 pm #1366185Douglas FrickBPL Member
My vote is to drop the ‘rank’ feature. Of course, my vote is low-ranked :)Nov 3, 2006 at 8:17 pm #1366191
Bill Cooper, 10/29 wrote:
This change was implemented 11/3/06. We’re testing code first before making it forum wide as part of your account settings, but you now have this available as a JS wrap.
/RJNov 22, 2006 at 11:07 am #1368031Mitchell KeilMember
@mitchellkeilLocale: Deep in the OC
A question back to BPL rankings. I know that this is a work in progress but I am getting confused by it. Beginning about two weeks ago I noticed that my ranking was 5.40 after having been below that for quite awhile then it dropped to 5.39 and now 5.20. What have I done to deserve this ignominy? (Just kidding, but this number has meaning — yes? So, why are you guys torturing us with it when we don’t know what it means?)Nov 22, 2006 at 11:17 am #1368037
It’s based on a random number generator influenced by readings from the eggs at the Global Consciousness Project at Princeton U.
We’re monitoring the rankings and trying to figure out their dynamics as well, before installing the next definition of the algorithm.Nov 22, 2006 at 11:33 am #1368043Eric NobleBPL Member
@ericnobleLocale: Colorado Rockies
I know it is too much to ask but I wish the quality of posts could play a large part in this number. I’ve seen instances where a number of so so posts seems to have raised the BPLRank of some new posters significantly. They are now higher than staff or those posting solid info for years.Nov 22, 2006 at 11:43 am #1368045
>> I’ve seen instances where a number of so so posts seems to have raised the BPLRank of some new posters significantly. They are now higher than staff or those posting solid info for years.
Agreed. That’s why the whole thing is still in ‘beta’.
Longevity of membership at BPL has not yet been factored into the system, but it certainly will be in the next iteration, as it should. Members have access to all the content which affords them greater opportunity to tap into other sources of information here when posting.Nov 22, 2006 at 11:55 am #1368047
Certainly there would be nothing wrong with having, and perhaps advisable to have, a “weighting factor” for Posters.
The “weighting factor” would be used in the calaculation of the “Ranking” together with other factors which BPL determines and rightly keeps “secret”.
For example, BPL Staff gets one high weighting factor value, and Industry “Heavy Hitters” (GVP, Ron Bell, Ron Moak, Dave Olsen, Henry Shires, etc. – sorry, if i missed mentioning any of y’all) also get a similarly high, more or less, weighting factor.
The “weighting factor”, for normal, run of the mill Forum Participants like myself would be lower, and, if it seems advisable, could increase with years of participation in the Forums – though i haven’t thought much about this last point to know if it’s a good suggestion.
Additionally, there might even so-called “expertise” areas for Forum Participants – enter three areas (more perhaps for BPL Staff and Industry “Heavy Hitters”) in one’s profile that one has some measure of “expertise” (i mean this in a liberal sense) and Posts on these subjects, based upon the Threads that the Posts are made in, would count more towards one’s Ranking than Posts in areas that a person doesn’t think they have as much to offer. Sure, more complexity, and it’s being left up to each participant to identify their supposed areas of expertise. In one sense this might work. I don’t Post much in the DIY Threads, but Bill Fornshell does and is quite knowledgeable in this area – surely, his posts in this area should be weighted far more heavily than my own congratulatory posts (this is another “wrinkle” to be ironed out) complimenting his work.
Clearly if anyone wants evidence that the Ranking should be taken with a grain of salt all they need do is look at DrJ’s rank and compare it with this Poster’s rank (in the case of this Poster, “rank” should perhaps have its other meaning!!!). Look, let’s face it, compared to DrJ, this poster is a negative number!
Rank, unlike a simple numerical account of the “# of posts” implies in the mind of some, some hierarchical level of knowlege or authority. Yet “Rank” is perhaps more useful for the implementation of any “bennies” that BPL has planned.
The more detailed (and complicated) the Ranking algorithm becomes, the more work it is to implement and shake out the “bugs” in it.
Anyways, just some thoughts.Nov 22, 2006 at 11:59 am #1368049Brett .Member
An engineer friend once gave me wise advice; if you aren’t going to use [that] data later, don’t collect it. And to test the ‘use’ part; shut off the flow of data and see who screams, if anybody.
So, is the BPL rank actually used for anything but impressing the girls behind the counter at REI? I mean.. if someone were to try that.. hypothetically speaking..Nov 22, 2006 at 12:08 pm #1368051
Brett, IIRC, DrJ has mentioned in the Forums, both a while ago and again much more recently, that BPL wants to reward Forum Participants with “bennies” (who know what they are? maybe add’l discounts in the BPL.com “store”??? Or, a chance to go trekking with DrJ in Montana or Yellowstone? So, some sort of “BPL Bucks” is in the works, if i’m not mistaken.
Sure, the Forums were great w/o Rankings, but now DrJ wants to give back a little to the Forum community if i’m not mistaken.
Anyways, i agree with you. No one will hear a peep from me if they magically go away (not that i think that they necessarily have to vanish – it’s all up to BPL to decide what to do.).
Oh, BTW, wise engineer friend you have there. I work with a lot of young engineers who have yet to learn that lesson your friend points out.
Also, do you really think that (if i weren’t married), a short, ugly, gruff old geezer like myself could really impress a young outdoorsy type of girl at REI with my BPL Forum ranking? Hmmm…it makes one think.Nov 22, 2006 at 2:12 pm #1368073Mitchell KeilMember
@mitchellkeilLocale: Deep in the OC
Jordan:”we’re monitoring the rankings and trying to figure out their dynamics as well, before installing the next definition of the algorithm.”
You’re trying to “figure out their dynamics as well.” Who developed this, the knomes of Zurich in secret in the dead of night? That you are not sure what is going on leaves me wondering about you guys. Does it make sense to run with something like this “live”?
I would fire my IT guy if he experimented with my system like this and then admitted he was not sure what the “dynamics” were of his “algorithm.” (But let’s give a shot, anyway!)
Please stop experimenting with us until you know what you are doing and why.Nov 22, 2006 at 2:38 pm #1368083
What I meant was this: I can’t predict user activity. I can only predict its impact on the rankings. We know the formula, I mean, it’s a formula! That’s not the issue. It’s the dynamic nature of user activity that is the issue.
Sorry for the confusion!
Question: “I would fire my IT guy if he experimented with my system like this and then admitted he was not sure what the “dynamics” were of his “algorithm.” (But let’s give a shot, anyway!)”
Answer: We do a lot of things here, be it on the website, or in the field with gear, that is … experimental without guaranteed outcomes. It keeps things fun. A Dilbert cartoon we are not.
The only thing we never mess with “experimentally” is the master user and order databases. Changes to those are always staged first and well tested before going live.Nov 22, 2006 at 5:09 pm #1368100Sam HaraldsonBPL Member
@sharaldsLocale: Gallatin Range
In my opinion I’m glad it’s more than a simple post count as with most PHPBB2 Forums. You tend to get quantity not quality in postings with that system.
I’ve always been impressed with the quality of posts here on BPL as well as how “on topic” the conversations tend to stay.Nov 22, 2006 at 8:50 pm #1368124Brett TuckerMember
@blister-freeLocale: Puertecito ruins
An algorithm that objectively measures quality…
Just imagine the quality of that algorithm!Nov 27, 2006 at 4:35 pm #1368519Richard MatthewsMember
@foodLocale: Colorado Rockies
You might want to change from “Rank” to “Participation Points”.
I suspect your lawyer would prefer that you not “rank” the forum advice.Nov 27, 2006 at 5:04 pm #1368523Kim SkaarupMember
@skaarupLocale: Cold, wet and windy Scandinavia
In a lot of reviews on everything one can often see a notice like “213 users found this information useful”. Now that could give you some points/higher rank. Not all our normal smalltalk.
Now my rank just increased. :-)Nov 27, 2006 at 7:46 pm #1368538Mike BarneyMember
@eaglembLocale: AZ, the Great Southwest!
This is a very classic, but straight forward problem that gets solved every day: You can see the output of a system, can guess what the inputs are (or they are from a finite and limited input set), and need (or want) to know how what the formula (algorithm) is, and how to duplicate it (Or possibly you know what the algorithm is and want to know what the (otherwise secret) inputs are.
What some have expressed a desire to do is to determine the relative weightings of the coefficients of various activities on this site: What do I have to do to get my rating to go up (or down, or stay at least 1/2 of pj’s :)
To determine the components (postings, time, reviews, etc) and their coefficients (weightings) of the ranking formula, you need to go through a little analytical exercise. Start with some free time and a sharp pencil or preferably a spreadsheet.
Next, collect individual user parametrics by pulling the participation information from users over time. (You can get a good static view by not including time as an element, as it appears as only a minor factor used in the rating equation.)
From there, it’s a simple matrix or simultaneous equations exercise to reverse engineer the ratings by mapping the postings in the various segments (forums, reader reviews, etc) vs. their ratings changes.
One can see there are more equations (user profiles vs. time) available than variables, so this becomes pretty straight forward.
So either crunch the numbers, put them in a spreadsheet, or get one of the matrix solving programs out there that can give you the the full equations w/ coefficients in a few milliseconds.
In addition to being active in postings, my quick back of the envelope analysis says Reader Reviews have a “significant weighting coefficient”, while time since last use is a consideration.
Wow, and to think I used to get paid to do this stuff. Sheesh! I think I’d rather go hiking!Nov 28, 2006 at 1:25 am #1368555
>>”1/2 of pj’s :)”
i’ve asked DrJ to do something about this. we definitely don’t want “pj’s” turning into a standard unit of measure for rank!
>>’Reader Reviews have a “significant weighting coefficient”‘
noticed that also. for any “rank” conscious individuals out there in Forum-land, try doing a few simple “Reader Reviews” and see what happens to your rank – at least until BPL reads Mike’s Post and adjusts the algorithm.
Oh, and Mike,… i like the wee touch of understatement, viz. “it’s a simple matrix or simultaneous equations exercise” – the operative word being “simple” – i also, like the implied “the proof is left to the student”.Nov 28, 2006 at 6:28 am #1368567Mike BarneyMember
@eaglembLocale: AZ, the Great Southwest!
But pj, you ARE the STANDARD!! (ok, have to put Bill F in there also)
Understatement? Actually, compared to crypto from a past life, this becomes a trivial matrix.
I would expect the real goal is to encourage more people to provide more input to improve the variety and depth of knowledge on the site.
Who knows, maybe Ryan will provide cupie dolls (ultralightweight versions of course) to pj’s and above :)
MikeBNov 28, 2006 at 9:32 am #1368573
>> maybe Ryan will provide cupie dolls (ultralightweight versions of course) to pj’s…
This is valuable input. Thank you for that. ;)
The algorithm is done, but not installed yet. We’re just waiting for the code developer to wrap it up. It will dramatically change current rankings.
What is in there now is pretty easy to decipher but it is a nonlinear matrix model, so you’ll have to invoke some regression analysis to really figure out. The tougher part is that nobody’s ranking is absolute: your bplrank is essentially a measure of your “contribution to bpl” that is normalized to the sum of all contributions by all bpl users.
When the new algorithm is installed there will be a nice description of the nature of the algorithm.
- You must be logged in to reply to this topic.
Our Community Posts are Moderated
Backpacking Light community posts are moderated and here to foster helpful and positive discussions about lightweight backpacking. Please be mindful of our values and boundaries and review our Community Guidelines prior to posting.