Not Logged In Log In   Sign Up   Points Leaders
Follow Us    1:00 AM

Message Forum - Read Message

Category: Suggest a GasBuddy improvement > Topics Add to favorite topics   Post new topicPost New Topic
Author Topic: A small idea to reduce wrong/false entries... Back to Topics
TheJediCharles

Veteran Author
Fort Worth

Posts:377
Points:66,285
Joined:Jul 2013
Message Posted: Sep 25, 2013 12:22:23 AM

I wondered about a simple technique that might remedy inaccurate entries, regardless if it's a simple goof or cheating for benefits. You guys might have already thought of it, and if so, sorry, and feel free to lock this one up as being redundant.

Is it possible that no entry actually appear or fully post (at least to anyone other than the reporter) until a second entry by a different reporter actually equals it? If it doesn't, the misfit, wrong data never appears to mislead anyone and is dumped. If the second agrees thus confirms the first, it then "releases" the first reporters entry and posts it until a third confirms the second and so on. New prices aren't a problem because it will be confirmed and end up appearing. A simple confirm-delay method.

The main drawback I can think of is the delay. But it might be a worthy tradeoff for truth in the numbers.

As usual, I'm probably missing another drawback, but I can't think of it.

What do you think?

[Edited by: TheJediCharles at 9/25/2013 12:23:51 AM EST]
REPLIES (newest first) Post a Reply
Profile Pic
Gas_Buddy
Champion Author Maryland

Posts:29,230
Points:3,490,960
Joined:Aug 2004
Message Posted: Oct 2, 2013 10:45:33 PM

I'm pretty much in agreement with Retired-Coastie's response regarding:

"I'm inclined to say that self-verification should not count; i.e., you can post a new price, but it would simply replace your previous price in the "needs to be verified" queue, rather than pushing it from the queue to the display. Otherwise I think we'd have a new problem of people posting everything twice."

There are several stations for which I'm the primary poster; I know that some other members post the prices for the station periodically, the same members, which leads me to think that they depend on me (at least based on Top Spotter numbers) for price posting for those stations. If I cannot self-verify, regardless of whether the price changes or remains the same, that means that there would be no posted price until one of the others posts the prices. My instinct is that they shop at the station based on my posting and then they verify the price when they go to get gas (I'm assuming they find my posting accurate and reliable). While I don't post prices for those stations every day, that (non-self-verification) means that the price probably won't be listed except by "chance opportunity".

I'm not advocating self-verification, but many of the members post a price in the morning and then again in the evening, whether the price changes or not. They're the ones the others are dependent upon. Especially when you recall that a significant number of members (I believe 15 percent or perhaps 25 percent) according to a recent poll either never or very infrequently post prices themselves.

Profile Pic
Retired-Coastie
Champion Author Arkansas

Posts:1,418
Points:472,860
Joined:Oct 2011
Message Posted: Oct 2, 2013 9:39:36 AM

"I'm inclined to say that self-verification should not count; i.e., you can post a new price, but it would simply replace your previous price in the "needs to be verified" queue, rather than pushing it from the queue to the display. Otherwise I think we'd have a new problem of people posting everything twice."

This would cause all three stations in my town to fade away in GB, as I know of a couple of people that use the app, but do not post. And, I have been the only one in my town to post for weeks if not months, with exception of the one post wonder driving through.
Profile Pic
kwzh
Champion Author San Jose

Posts:22,816
Points:4,133,755
Joined:Jul 2001
Message Posted: Sep 30, 2013 4:36:40 AM

Understood. The reason we (or I, at least) have been considering "how" it's wrong is because it affects the analysis. (Though even if the math is right, it's not worth much unless we have some solid figures to feed into it.)

> (optimally by a different poster)

I'm inclined to say that self-verification should not count; i.e., you can post a new price, but it would simply replace your previous price in the "needs to be verified" queue, rather than pushing it from the queue to the display. Otherwise I think we'd have a new problem of people posting everything twice.
Profile Pic
TheJediCharles
Veteran Author Fort Worth

Posts:377
Points:66,285
Joined:Jul 2013
Message Posted: Sep 30, 2013 2:20:28 AM

Fine. But you two appeared to be discussing only the matter of "how" a wrong entry was wrong, and i was responding to that.

Also, it's only "throwing the baby out with the bathwater" if it in fact is. The discussion is the exploration of if it is or not, not that it simply is a given, because it isn't.
Profile Pic
TxJeans
Champion Author Tampa

Posts:6,797
Points:707,000
Joined:May 2004
Message Posted: Sep 29, 2013 2:34:30 PM

<<<<It doesn't matter "why" the price was wrong. The idea is, if the price doesn't agree with the following entry (optimally by a different poster) it doesn't post.>>>>>

Yes, but the solution throws out the baby with the bath water when we have too much lag in the prices showing up. Any increase or decrease would languish until verified, and if the price was wrong, and an "arm chair" poster "verifies it" it becomes a wrong POSTED price until two honest posters post the correct price.

The retroactive analysis MIGHT be useful in determining those that post fraudulent prices or have an abnormally high fat-finger error.
Profile Pic
TheJediCharles
Veteran Author Fort Worth

Posts:377
Points:66,285
Joined:Jul 2013
Message Posted: Sep 29, 2013 1:46:08 PM

It doesn't matter "why" the price was wrong. The idea is, if the price doesn't agree with the following entry (optimally by a different poster) it doesn't post.

[Edited by: TheJediCharles at 9/29/2013 1:46:47 PM EST]
Profile Pic
TxJeans
Champion Author Tampa

Posts:6,797
Points:707,000
Joined:May 2004
Message Posted: Sep 29, 2013 9:34:47 AM

I think there are more "point grabbing wrong" prices than "fat fingered" prices. so, I don't think it would work in real-time, but it might help with an analysis of trustworthiness of posters. Even if fat fingered, if the fat fingering occurs at a high level, it might be worth letting the poster know so they can slow down and be more careful with their entry -- or might provide clues to how to improve price entry if the errors are more often seen from certain entry modes/devices.
Profile Pic
kwzh
Champion Author San Jose

Posts:22,816
Points:4,133,755
Joined:Jul 2001
Message Posted: Sep 29, 2013 7:13:16 AM

TxJeans writes,
> I don't think your math proves that the proposal would improve accuracy...except in the case that someone types an incorrect price.

That's the only case I was considering, in the comment that you quoted; and my point was to show that it's not ONLY a warm-fuzzy thing, because SOMETIMES it's an improvement over the status quo.

(Also, the math was using the model that we only have genuine spotters and price echoers, while the improved-accuracy argument was assuming we have genuine spotters and honest typos. Two different models; separate discussions.)

> do you now support the delay and indicator or not?

Hmm... I think that the general principle is sound (if 9 people report a price of 4.51 and one person reports a price of 5.41, with that person being neither first nor last chronologically, then the outlier is probably wrong). I'd keep it in mind for future use -- possibly for determining trustworthiness of a member! -- but I guess I've allowed myself to be convinced that the stations aren't currently reported frequently enough for it to be a win.
Profile Pic
TxJeans
Champion Author Tampa

Posts:6,797
Points:707,000
Joined:May 2004
Message Posted: Sep 28, 2013 7:29:19 AM

So, kwzh, do you now support the delay and indicator or not?

"And the proposal would indeed improve accuracy, rather than just being warm fuzzies, at least in the case where someone accidentally types an incorrect price -- it dies for lack of verification, rather than overwriting the correct price. "

I don't think your math proves that the proposal would improve accuracy...except in the case that someone types an incorrect price. At least not in the "Speedway" world where they have lots of spikes and in rural areas where posting may not have as many members posting the same station.

Around me there are plenty of stations that are not reported multiple times a day or even every day. As much as I believe the APP have contributed to price inaccuracies seen by fuel grades being reported that are not there, I still say this fix would break more than it would fix.

Again, that is MY opinion and I have NO problem with you having your opinion, but just trying to understand it better. It is through discussions that minds (at least some) might change.
Profile Pic
kwzh
Champion Author San Jose

Posts:22,816
Points:4,133,755
Joined:Jul 2001
Message Posted: Sep 28, 2013 2:18:43 AM

Scrapheap writes,
> So does the person who posted the original change in price get his price reported with him getting credit for the report on the price pages or does the person who "verifies" the changed price.

The person posting the original change would get credit. But, when a *third* person posts the same price, then the *second* person's has been confirmed, and it will overwrite the first person's at that time. (That's how I read what the OP states.)
_____

kwzh writes,
>> My interpretation of the "10 cents" comment was that it was just an example, and it would be equally true for a change of 2 cents, say.

TxJeans writes,
> [...] Or are you talking about rspiano's example - which was obviously just an example - but obviously, a 20 cent difference held over time would be a bigger surprise than 2 cents.

Yes, rspiano's example is what I was referring to.

> What would be equally true of 2 cent changes? That they are "uncommon" or that the system would not accept?

That, if the proposal were to be adopted, this price change would not accepted until confirmed by another member.

> Are you changing your opinion from the 25th?

I didn't state an opinion on the 25th. I did some math and announced what a certain model implied about the delay. Whether the increase in lag time is acceptable (to me) or not would depend on the frequencies involved, and I chose not to guess any values for them.
Profile Pic
Gas_Buddy
Champion Author Maryland

Posts:29,230
Points:3,490,960
Joined:Aug 2004
Message Posted: Sep 27, 2013 7:49:34 PM

"Suppose the confirming number could be simply a visual indicator (like the number is a green color as opposed to black, or a bullet point alongside) that means "price twice confirmed". THAT way, the first, unconfirmed (perhaps wrong) update lacks that confirmation, so the number is at least there to be seen for what it's worth, but the confirmed prices have the additional credibility we're looking for in a site so regularly harmed by wrong entries>

Personally I don't see people remembering that a bullet or a different color would be indicative of "this price might not be right because it's not "twice confirmed" or that "I might not want to shop there because it's not price confirmed" or "Oh yeah, that color means I can buy gas at a confirmed price (unless, of course, the price has changed again and no one twice confirmed it."

Too confusing.

And, do you really want "twice confirmed"? Twice confirmed means that the price was posted and someone confirmed it, and then someone else confirmed it. Twice posting the same price means that it was confirmed (confirmed one time, not twice confirmed). Do you really want a delay in price posting until two members "confirm" what the first member posted?

Profile Pic
TxJeans
Champion Author Tampa

Posts:6,797
Points:707,000
Joined:May 2004
Message Posted: Sep 27, 2013 6:18:37 PM

I tend to base the validity of prices by the name posted next to it...around here I tend to know which posters are serious gasbuddy supporters and which ones are playing with the app and will soon fade away to be replace by the next newcomer trying out a new toy...

Profile Pic
TheJediCharles
Veteran Author Fort Worth

Posts:377
Points:66,285
Joined:Jul 2013
Message Posted: Sep 27, 2013 4:03:06 PM

"I alreeady thought of that"

Too late. I already Al Gored your internet invention.
Profile Pic
Scrapheap
Champion Author Virginia

Posts:16,904
Points:2,813,785
Joined:Sep 2006
Message Posted: Sep 27, 2013 2:07:20 PM

CK - Trooper207 was suggesting a 1 mile radius limit. He doesn't believe it currently exists.
Profile Pic
CampKohler
Champion Author Sacramento

Posts:12,335
Points:1,943,315
Joined:May 2007
Message Posted: Sep 27, 2013 1:58:38 PM

JediChuck: I alreeady thought of that (see "...a repeated price could turn on an indicator [change some appearance of the price display, etc.]...." on the 25th), so it is I that deserves the 5% commission of the system revenue and not you. :-)

Trouper207: The one-mile-radius-posting-limit for the mobile app is an absolute myth (or it is failing today). I tested it by re-posting the prices of a station two miles away at my destination, but bumped the premium by 1 cent. When I arrived 15 mins later, the premium price was 1 cent too high and my nickname was on all the prices. I then reported the correct prices. It would be a mistake preventer if you accidentally reported the wrong station (of the same brand) that was over a mile away and it rejected your report, but it doesn't. Now, is something broken, or was that rule just another GasBuddy urban legend?

[Edited by: CampKohler at 9/27/2013 2:01:04 PM EST]
Profile Pic
TheJediCharles
Veteran Author Fort Worth

Posts:377
Points:66,285
Joined:Jul 2013
Message Posted: Sep 27, 2013 11:27:38 AM

OH WAIT!

I just thought of something. A different way to use "confirming" in a less revolutionary manner. Suppose the confirming number could be simply a visual indicator (like the number is a green color as opposed to black, or a bullet point alongside) that means "price twice confirmed". THAT way, the first, unconfirmed (perhaps wrong) update lacks that confirmation, so the number is at least there to be seen for what it's worth, but the confirmed prices have the additional credibility we're looking for in a site so regularly harmed by wrong entries.

How about THAT? :D Eh? Hm?

[Edited by: TheJediCharles at 9/27/2013 11:29:36 AM EST]
Profile Pic
TheJediCharles
Veteran Author Fort Worth

Posts:377
Points:66,285
Joined:Jul 2013
Message Posted: Sep 27, 2013 11:02:04 AM

There's been several responders that are pointing out the time limit before prices drop off as being too old as a limit to this idea. Now, obviously there's a delay, but the suggestion that the drop-off time limit isn't able to be adjusted in conjunction with this confirming idea is not true. It can also be modified and lengthened as the numbers being doubly reliable (since the clock doesn't start since the 2nd confirmation anyway).
Profile Pic
TxJeans
Champion Author Tampa

Posts:6,797
Points:707,000
Joined:May 2004
Message Posted: Sep 27, 2013 7:14:13 AM

kwzh
"My interpretation of the "10 cents" comment was that it was just an example, and it would be equally true for a change of 2 cents, say."

CK: "10 cent changes are not common except when someone has mistakenly reported a credit price instead of a cash price"

Or are you talking about rspiano's example - which was obviously just an example - but obviously, a 20 cent difference held over time would be a bigger surprise than 2 cents.

What would be equally true of 2 cent changes? That they are "uncommon" or that the system would not accept?

"And the proposal would indeed improve accuracy, rather than just being warm fuzzies, at least in the case where someone accidentally types an incorrect price -- it dies for lack of verification, rather than overwriting the correct price. "

Yes, this would improve the accuracy of that one type of error, but at what expense? And at what expense of other accuracies? Are you changing your opinion from the 25th? Or can you explain how CK's post has solved the issues you stated on the 25th? Seriously, I don't see how it would work based on what has been posted so far. Maybe you an help explain how it would be deployed successfully and not do more damage than it solves?

Profile Pic
scoutmaster
Champion Author Pittsburgh

Posts:94,462
Points:3,708,615
Joined:Mar 2003
Message Posted: Sep 27, 2013 7:07:31 AM

"It would only take one verification to prove a price is not in error"

Yup and what if that verification is for an incorrect price?
Profile Pic
scoutmaster
Champion Author Pittsburgh

Posts:94,462
Points:3,708,615
Joined:Mar 2003
Message Posted: Sep 27, 2013 7:06:29 AM

"And the proposal would indeed improve accuracy, rather than just being warm fuzzies, at least in the case where someone accidentally types an incorrect price -- it dies for lack of verification, rather than overwriting the correct price."

This proposal MIGHT improve accuracy. Stating it would INDEED improve accuracy is being unrealistic.

I personally think this idea doesn't float. There could be considerable gaps between a price being posted for a station which would prevent a price from appearing on the site. Plus, if the price changes between the 1st and 2nd post there will need to be a 3rd post to verify the new price. Too much of a delay.
Profile Pic
Scrapheap
Champion Author Virginia

Posts:16,904
Points:2,813,785
Joined:Sep 2006
Message Posted: Sep 27, 2013 6:40:42 AM

So does the person who posted the original change in price get his price reported with him getting credit for the report on the price pages or does the person who "verifies" the changed price. It should go to the latest report, which will reduce people's willingness to report changed prices.

A Safeway near me recently went to dual pricing. GasBuddy Direct and I report the cash price. Just about everyone else reports the credit price. How is that situation handled?
Profile Pic
kwzh
Champion Author San Jose

Posts:22,816
Points:4,133,755
Joined:Jul 2001
Message Posted: Sep 27, 2013 1:32:26 AM

My interpretation of the "10 cents" comment was that it was just an example, and it would be equally true for a change of 2 cents, say.

And the proposal would indeed improve accuracy, rather than just being warm fuzzies, at least in the case where someone accidentally types an incorrect price -- it dies for lack of verification, rather than overwriting the correct price.
Profile Pic
Gas_Buddy
Champion Author Maryland

Posts:29,230
Points:3,490,960
Joined:Aug 2004
Message Posted: Sep 26, 2013 7:48:27 PM

Trooper207:

"Have seen someone posting prices to multiple stations within minutes when the stations are physically 10 or more miles apart . Very frustrating."

Not really. Not really frustrating that is. Not every Gas Buddy member posts gas prices on his/her mobile device and not every Gas Buddy member posts gas prices immediately after seeing the price as they drive by the gas station. Until relatively recently there wasn't a mobile app and members posted prices from their laptop or from their desktop computer after they arrived at work or home (for example). Many members still don't post prices until after they stopped driving and arrived where they felt safe to post gas prices or when they felt they had the time to post gas prices. Posting gas prices for gas stations 10 miles apart isn't that questionable. It's not uncommon for me to post gas prices, for example where I live in Maryland and for where I commute to in D.C., some 25 miles apart, nor is it uncommon for me to post gas prices in the northern Florida/southern Georgia area from my starting point and my arrival point, approximately 20-25 miles apart.
Profile Pic
TxJeans
Champion Author Tampa

Posts:6,797
Points:707,000
Joined:May 2004
Message Posted: Sep 26, 2013 7:27:43 PM

CK: "I will say 10 cent changes are not common"

Not on a day to day, but I have often seen 10 cent jumps, and I bet it is not that rare up in the "Speedway" area where RG is located.

CK: "If a price of $3.89 is correct, it is correct no matter how many times it is reported. The system (as it stands) doesn't care how many times a particular price was reported; the price can't get any more closer to $3.89, that is, it can't get any more correct. Say there was a counter that kept track of how many times* $3.89 was reported. If the count was falsely bumped up 20, the humans would look upon the price in a more happy light (warmer and fuzzier, if you will), but the price would no less or more correct."

Huh? $3.89 can be reported 20 times correctly, but if the price changes and someone that did not see the change at the station copied the $3.89, it is no longer "correct". A counter for accuracy would only be valid if there were multiple price postings in a very short time. In most cases that is not the case.

CK: '*It would only take one verification to prove a price is not in error, unless of course, members are considered so unreliable that two reporting the same price is not sufficient and some sort of voting would be required to reject a one price different than the others. But that would entail a big delay to wait for all the votes to come in and get smoothed out into some kind of a curve."

How would you know if the verification was truly a verification or a copy over by some lazy poster looking for points? You don't. Having a vote and "smoothed curve" is not going to solve anything.

The best bet for improvement in accuracy would be:
1 - making the price entry fields tied to the MSL
2 - strict review of reported cheaters and limit their ability to post

The first one is probably the most likely and hopefully in the works.
The second one is problematic in verifying the poster is an abuser rather than a fat fingered mistake. Mistakes will always happen no matter how careful people are when posting prices on small devices or from lists that they capture to enter when they get home.

Profile Pic
Trooper207
Rookie Author Maryland

Posts:2
Points:514,170
Joined:Jul 2012
Message Posted: Sep 26, 2013 7:10:43 PM

How about if the location service on your phone is more than 1 mile away, then the price won't post, as you are too far to really report it. Have seen someone posting prices to multiple stations within minutes when the stations are physically 10 or more miles apart . Very frustrating.
Profile Pic
TheJediCharles
Veteran Author Fort Worth

Posts:377
Points:66,285
Joined:Jul 2013
Message Posted: Sep 26, 2013 5:10:05 PM

"Maybe the solutions isn't outside the box, but something inside the box, something more simple."

As Einstein said, "everything should be made as simple as possible, but not simpler."

So, it's only a question of if what we have now is simple enough, or too simple. Depending on who you ask, some might say the number of wrong entries might suggest it's too simple.
Profile Pic
CampKohler
Champion Author Sacramento

Posts:12,335
Points:1,943,315
Joined:May 2007
Message Posted: Sep 26, 2013 4:03:03 PM

GB: For "warm and fuzzies," read "to make humans feel better." If a price of $3.89 is correct, it is correct no matter how many times it is reported. The system (as it stands) doesn't care how many times a particular price was reported; the price can't get any more closer to $3.89, that is, it can't get any more correct. Say there was a counter that kept track of how many times* $3.89 was reported. If the count was falsely bumped up 20, the humans would look upon the price in a more happy light (warmer and fuzzier, if you will), but the price would no less or more correct.

----
*It would only take one verification to prove a price is not in error, unless of course, members are considered so unreliable that two reporting the same price is not sufficient and some sort of voting would be required to reject a one price different than the others. But that would entail a big delay to wait for all the votes to come in and get smoothed out into some kind of a curve.
Profile Pic
Gas_Buddy
Champion Author Maryland

Posts:29,230
Points:3,490,960
Joined:Aug 2004
Message Posted: Sep 26, 2013 4:00:36 PM

"At least I try to get outside the box in so doing.

I don't disagree with TheJediCharles that outside the box solutions are not feasilbe or might not work but I've found some or many of the out of the box suggestions or solutions to be overly dynamic or somewhat convoluted whereas the website tries to operate on simplicity and immediacy. Maybe the solutions isn't outside the box, but something inside the box, something more simple.
Profile Pic
TheJediCharles
Veteran Author Fort Worth

Posts:377
Points:66,285
Joined:Jul 2013
Message Posted: Sep 26, 2013 3:48:01 PM

Yep, this is all thoughtful stuff. I'm guessing the silver bullet to solve inaccuracies is just pie in the sky, but I'm simply stuck in a brain that refuses to give up finding it. At least I try to get outside the box in so doing. I still have this gut feeling that in the right hands this could be the germ of an idea that can be modified into something that does work.

But when I say "work" I don't mean without some form of drawback. In the end, this is a battle of competing priorities, and even not changing a thing, what we have now, albeit good, is still flawed. Any set of benefits chosen comes with the choosing against another. Compound that with the differing values of all the users and nothing will ever be perfect. Some lean towards "more numbers of less perfect entries" and others "less numbers of more perfect entries" and I guess that is unfixable.

Thanks for exploring it anywho.

[Edited by: TheJediCharles at 9/26/2013 3:49:58 PM EST]
Profile Pic
CampKohler
Champion Author Sacramento

Posts:12,335
Points:1,943,315
Joined:May 2007
Message Posted: Sep 26, 2013 3:46:38 PM

rspiano: Are you saying that the system will not now accept 10 cent or greater changes without verifying reports? Or are you suggesting that?

I will say 10 cent changes are not common except when someone has mistakenly reported a credit price instead of a cash price. I haven't personally seen any of my prices rejected for price jumps. However I noticed a station that sold diesel for $4.33 yesterday sells it for $4.09 today, but I wasn't the one who reported the jump.
Profile Pic
Gas_Buddy
Champion Author Maryland

Posts:29,230
Points:3,490,960
Joined:Aug 2004
Message Posted: Sep 26, 2013 1:49:44 PM

In addition to prices being (potentially) stale or simply not updated (because of the delay in waiting for a confirmation post, nothing would prohibit the same member (who you think is intentionally posting false prices) from simply reposting the false price, in essence, confirming his/her own post.
Profile Pic
rspiano
Champion Author Tucson

Posts:11,013
Points:2,257,230
Joined:May 2008
Message Posted: Sep 26, 2013 11:15:03 AM

If I post a station which increased their prices by 10 cents over the previous post, my update will not take effect until someone else "verifies" it, which for some stations might take hours or the better part of a day. In the meantime, someone who sees the previous (now incorrect) price might drive to that station hoping to get a bargain, and see that the price jumped 10 cents. That person may be one of the many users who use GasBuddy for information but do not post prices on their own, therefore the incorrect price will stay on the board.

Conversely, if I post a station which *decreased* their prices by 10 cents, that bargain hunter will not know about the decrease until someone else verifies the new price. The incorrect price will most likely stay on the board longer, as those who see the higher price (including the bargain hunter) for that station will probably fill up elsewhere.
Profile Pic
scoutmaster
Champion Author Pittsburgh

Posts:94,462
Points:3,708,615
Joined:Mar 2003
Message Posted: Sep 26, 2013 6:13:07 AM

Warm and fuzzy? I don't see that either Gas Buddy. I don't like the suggestion but I don't think the OP was going for a warm and fuzzy.
Profile Pic
Gas_Buddy
Champion Author Maryland

Posts:29,230
Points:3,490,960
Joined:Aug 2004
Message Posted: Sep 25, 2013 11:10:52 PM

"Either of these would do not much more than give users "the warm and fuzzies," but that's all this suggestion was intended to do."

While I disagree with the suggestion, I didn't get the impression that TheJediCharles was simply trying to give users "the warm and fuzzies". I got the distinct impression TheJediCharles was trying to insure viewers saw more accurate prices.
Profile Pic
CampKohler
Champion Author Sacramento

Posts:12,335
Points:1,943,315
Joined:May 2007
Message Posted: Sep 25, 2013 5:04:26 PM

This suggestion has been added to the Suggestion Tracking List as a new subject.

The delay problem for less-frequently-updated stations would be a killer; if such a station was routinely reported only by one member, it would never get updated. Even for stations that were reported, say, twice a day, the lag would be significant. In summary, I believe it would be too much of a cost to users to trade currency for validity.

This suggestion is "before the fact" (of reporting), if you will, but it could be made "after the fact." Reports that agree with previous reports from other members could be used as a validity check; a repeated price could turn on an indicator (change some appearance of the price display, etc.) so that it will be apparent that more than one member has reported the same price, or increment a counter in the price display with each repeat of the price. Either of these would do not much more than give users "the warm and fuzzies," but that's all this suggestion was intended to do.
Profile Pic
Gas_Buddy
Champion Author Maryland

Posts:29,230
Points:3,490,960
Joined:Aug 2004
Message Posted: Sep 25, 2013 3:52:05 PM

I'm not sure I agree with kwzh's figures but it's likely that for the more frequently posted stations the time delay would be relatively insignificant. But it's the unknowns that are significant. For example, while some prices may be viewed frequently by commuters, they may not be updated by a significant number of members. that means that current pricing will be delayed and prices viewed more likely to be inaccurate even though the actual current price has been posted (it's in limbo waiting confirmation).

Older, less traveled stations, backroads and/or off-the highway may be delayed by as much as a day or longer through inaction.

Because I don't see false pricing to be a significant problem in the areas I post prices (fat fingers notwithstanding), I would rather have real time posting than delayed confirmation because, in part, many people depend on prices along the way but don't necessarily post prices for those areas, though they do post for their immediate home and/or work areas.
Profile Pic
Retired-Coastie
Champion Author Arkansas

Posts:1,418
Points:472,860
Joined:Oct 2011
Message Posted: Sep 25, 2013 8:21:16 AM

kwzh - the math looks scary.

TJC - Glad to see you chiming in with a suggestion.
Profile Pic
TxJeans
Champion Author Tampa

Posts:6,797
Points:707,000
Joined:May 2004
Message Posted: Sep 25, 2013 6:51:58 AM

There are two stations I report most days. Someone other than me may report every few days. The prices would be constantly stale. So, these would for the most part disappear or have old prices - and I am not in a rural area where it would be compounded with less membership. These two stations are at a busy corner.I don't know why they don't get reported more often, but they don't. When I see their prices change, I know my cheaper stations will be also changing in the same direction.

I agree with the others that it is not an idea that will improve the data.
Profile Pic
scoutmaster
Champion Author Pittsburgh

Posts:94,462
Points:3,708,615
Joined:Mar 2003
Message Posted: Sep 25, 2013 6:18:16 AM

Anything that delays the posting of prices is not a good idea. I don't think we will ever rid this site of incorrect prices. The only thing that will eventually happen is to eliminate the ability to post a price for a grade a station doesn't sell. This will be a huge benefit when it goes live.
Profile Pic
kwzh
Champion Author San Jose

Posts:22,816
Points:4,133,755
Joined:Jul 2001
Message Posted: Sep 25, 2013 4:32:51 AM

Interesting idea. Let's see how bad the delay problem would be.

The time between an actual price change and the new price appearing on the board is currently the time it takes for one member to report it. With this proposed change, that time would double, as we wait for a second member to come along and confirm it. If we need to also deal with the clock being reset by armchair updaters who are reporting the still-visible old price, then instead of a factor of 2, we have a factor of (2 + b/g), where b and g are the rates at which bad and good reports arrive (assuming a simple Poisson process model). To be precise, the delay time would increase from 1/g to (2 + b/g)/g.


[Edited by: kwzh at 9/25/2013 4:33:33 AM EST]
Profile Pic
maxstar
Champion Author Chicago

Posts:21,503
Points:1,266,400
Joined:Feb 2011
Message Posted: Sep 25, 2013 1:08:56 AM

How would it work if the second prices reflects a legitimate price increase or decrease.

Also some stations may not have prices updated very often. How would that process work in areas where stations are not updated every day.
Post a reply Back to Topics