BrettM wrote:

This is going to take some head scratching.

The numbers for 11/7/07 have been posted at participating hood imagers, including the Pellet Points hood, to induce more head scratching and shampoo sales.

_________________
OpenUru.org: An Uru Project Resource Site : Twitter : Perfect Speed Is Being There.

I should learn to be more wary of thoughts that come to me in the early morning pre-caffeine.

If the daily average is the number of points for the day divided by the number of explorers dropping, then how could there be spikes due to variation in participation? Assuming that almost everyone dropping pellets these days is using a high-scoring recipe, it wouldn't matter then whether 10 people dropped on a particular day or 100 or 1000: the average would be about the same.

If it really is the average number of points per explorer, then there must be a wide variation in the recipes being used. In which case the solution to smoothing out the line is simply education. Make sure that one of these recipes is widely known and used.

The question is how can we determine which is the case? Is the average varying because of the varying number of participants, or is it varying because of differences in the recipe? If the former, then we have to figure out just what the heck is actually being averaged.

Honestly, you'd think Laxman was trying to set up a self-fulfilling prophesy. Giving us numbers without the information needed to interpret them properly so later he can say "See! I knew you'd misuse them!"

_________________

*SLMW 1.0* No animals were harmed in the production of this message.

BrettM wrote:

If the daily average is the number of points for the day divided by the number of explorers dropping, then how could there be spikes due to variation in participation? Assuming that almost everyone dropping pellets these days is using a high-scoring recipe, it wouldn't matter then whether 10 people dropped on a particular day or 100 or 1000: the average would be about the same.

You raise an excellent point, though it does give one crucial other piece of info: "how many drops/day is each explorer making"? Which can also blossom into "how many sacrificial avatars are also making pellets for this explorer to drop," lol.

Victor is going well out of the way to make very sure we can't get pure, raw numbers to play with.

_________________
"I visited Esher's lab and all I got was this lousy t-shirt."
VidRoth -- KI#50637

BrettM wrote:

I should learn to be more wary of thoughts that come to me in the early morning pre-caffeine.

If the daily average is the number of points for the day divided by the number of explorers dropping, then how could there be spikes due to variation in participation? Assuming that almost everyone dropping pellets these days is using a high-scoring recipe, it wouldn't matter then whether 10 people dropped on a particular day or 100 or 1000: the average would be about the same.

I think Laxman just put the wrong lable on the number... I think we should assume that the daily average is just the total amount of added lake points of that day. Otherwise it just wouldn't make any sense.
I think Laxman was trying to say that the daily number is not the total amount of lake points, but the amount of lake points that have been added that day. He used the term "average" to say that, which is of course wrong.
That's what I think and what I think to be the most likely.

The weekly average is the average of the lake points of the 7 latest days. So when we've got a whole week of daily data, we'll even be able to calculate a weekly average ourselves. The total average is just the amount of lake points so far per day, we won't be able to calculate that ourselves, because then we need the data from every day.

Erik wrote:

The weekly average is the average of the lake points of the 7 latest days. So when we've got a whole week of daily data, we'll even be able to calculate a weekly average ourselves. The total average is just the amount of lake points so far per day, we won't be able to calculate that ourselves, because then we need the data from every day.

Looking at the numbers now that assumption seems to be correct.

The total number seems to be slowly decreasing (which makes sense, a lot more pellets were made when Er'cana was released which results in a high average). So if it's affected by the daily value you could calculate the number of days it has been measured and use it to calculate the total average.

(Get that today is day 150, can someone check? I fear I'm just fooling around with numbers again, heh. )

_________________
KI: 59666
http://www.uruobsession.com/

I have conducted some preliminary analyses of the data and the results are stunning. Because of the embargo imposed by JWP on the release of the data, however, and because it is highly probable that JWP's association with Laxman has given JWP certain resources for assuring compliance (read: ninja assassins), I am hesitant to publish my results.

_________________
Grand Master
Guild of Archivists

I had an idea... What if the Daily average is the total number of points divided by the number of days Er'cana has been available? The Weekly average then being the total divided by the number of weeks, etc.

_________________
-Whilyam

Whilyam wrote:

I had an idea... What if the Daily average is the total number of points divided by the number of days Er'cana has been available? The Weekly average then being the total divided by the number of weeks, etc.

Wouldn't that mean that the weekly average would always be exactly 7 times the daily average? Which would mean that they would be identical when expressed as fractions of the 100% value?

If the number of weeks used is constrained to be an integer, the above would still be the case every 7th day.

It appears the experiment will continue. I have received encouraging word of satisfaction.

The pellet numbers for November 12, 2007 are now posted to the imagers in the Pellet Points and other participating hoods.

_________________
OpenUru.org: An Uru Project Resource Site : Twitter : Perfect Speed Is Being There.

I think the "daily average" is just an error in the graph... After all, JWPlatt doesn't even get a daily average, but a value, a weekly average, and a total average. I still think that:
value a.k.a. "daily average" = added lake points during a day
weekly average = (values of the last 7 days) / 7
total average = (all values) / days since Er'cana release

And it's great to hear that the experiment is continuing!

Erik wrote:

I think the "daily average" is just an error in the graph...

You could be right, though I really don't want to believe it. Laxman is a trained engineer with a great deal of experience on top of impressive credentials. I can believe he is not well versed in graphical representation of statistical data (so few are!), however it is almost beyond belief that a man with his education and experience would not be crystal clear on the difference between a "total" and an "average".

_________________

*SLMW 1.0* No animals were harmed in the production of this message.

Perhaps JWP could ask what it means.

Figure it has to be (assuming Laxman means what he says, "average") either the total for the day divided by the number of explorers, or the number of hours.

_________________
-Whilyam

Hmm, well, what if it's an average of the values of the pellets divided by the number of pellets, rather than the number of explorers? I haven't messed with the pellet system too much, but the pellets have specific point values right? And when you turn them in it may be looking at the number of pellets you've made rather than treating all of the pellet points as a single value. So if everyone used exactly the same pellet formula, you'd get a number equal to a single pellet. If it *does* work like that... then there's no functional difference between turning in 100 pellets and turning in 1... as long as everyone were turning in the same pellet. If people aren't, then flooding the lake with better pellets would act to diminsh the less optimal ones to a greater degree.

The problem with that theory is that it doesn't really allow for spikes... unless they're measuring more than one value or the pellet point contributions are very inconsistent... Maybe, it works that we'd need to maintain an even amount of contribution and depending on the average amount of pellet points dumped into the lake, the lake improves by this much as a result. The percentage value we see then would be the pellet point total(or average) adjusted by the consistency of the input.

BrettM wrote:

--however it is almost beyond belief that a man with his education and experience would not be crystal clear on the difference between a "total" and an "average".

You never know. Even English professors say "hopefully".

_________________
-Whilyam