| Author |
Message |
|
|
Post subject:
 Posted: 03.30.2006 16:22
|
|
Joined: 15.03.2006 Posts: 9
Location: Toronto, ON, Canada
|
|
Quote:
I've got charts for both of these (because, see, I'm a geek). I'll try to post them later tonight.
Don't be so hard on yourself. You're a dork, at worst.
er...at best?
[edit: where'd the rest of my comment go? no matter]
Being a fellow of the dorkhood, I'd be very interested in the data myself.
[further edit:]
Have you run any data regarding a) number of entries and b) number of votes in relation to greenlight time? |
_________________
http://www.kylemckeown.com/overlooked
|
| |
|
|
|
 |
|
|
|
Post subject:
Posted: 03.30.2006 17:46
|
|
Moderator
Joined: 29.12.2005 Posts: 994
Location: Austin TX
|
|
IMeant2DoThat wrote: Being a fellow of the dorkhood, I'd be very interested in the data myself.
Heh
Finishers v Time from Greenlight
Vote Getters v Time from Boobies
Note on that last one - it only shows 6 contests. The correlation (r) is only for the top 40 vote getters. For this sample set, the hypothesis ("Early posting garners more votes") is likely to be false. The "p" value is a measure of the confidence in the correlation. Lower p's mean that there is less likely to be an error. Statisticians think that a p < 0.05 is a good value (if I recall my stats classes correctly). |
|
|
| |
|
|
|
|
|
|
|
Post subject:
Posted: 03.30.2006 20:12
|
|
Joined: 06.01.2006 Posts: 52
Location: Centrally Isolated
|
|
schnee wrote: For this sample set, the hypothesis ("Early posting garners more votes") is likely to be false. The "p" value is a measure of the confidence in the correlation. Lower p's mean that there is less likely to be an error. Statisticians think that a p < 0.05 is a good value (if I recall my stats classes correctly).
Could it be as simple as "early posts are near the top of the page and therefore everybody sees them"? |
_________________
//Share & Enjoy
|
| |
|
|
|
|
|
|
|
Post subject:
Posted: 03.30.2006 21:51
|
|
Moderator
Joined: 29.12.2005 Posts: 994
Location: Austin TX
|
|
Encyclopedia Galactica wrote: Could it be as simple as "early posts are near the top of the page and therefore everybody sees them"?
That, and early posts have more time to accumulate votes. The early post advantage is the hypothesis. TotalFark provides the opportunity for an early post. Fark could attempt to reduce that advantage by taking certain steps, but the load on the servers would be large and why would Fark, Inc. reduce the value of TF?
I'm actually pretty encouraged that the hypothesis is not strongly supported by the analysis (p-values are "large"). In my natural state, I'm a 'liter. I like to think I have some chance...
(note that as the analysis is run over all entries that garnered votes, in every contest, p approaches zero which means that there is a definite early post advantage. In this case, however, the analysis include entries with only one or two votes)
I'm guessing that there is a definite and strong correlation between TF'itude and landing in the top-three, but I haven't tested that (yet).
(Great Noodle, I really am a dork) |
|
|
| |
|
|
|
|
|
|
|
Post subject:
Posted: 03.31.2006 02:51
|
|
Joined: 29.03.2006 Posts: 2
Location: Bay Area, CA
|
|
I realize that I'm getting in here a little bit late to this covo, but did either of you notice that the full fark thread is over 14MB? For anyone, a liter or TFer on dial up that's a long download time. A lot of people probably aren't waiting for the photos at the bottom to load.
There are over 50 images that are 100+Kb!
That's my two cents worth here anyways. |
_________________
"I am a leaf on the wind; watch how I soar"
|
| |
|
|
|
|
|
|
|
Post subject:
Posted: 03.31.2006 04:31
|
|
|
|
Encyclopedia Galactica wrote: schnee wrote: For this sample set, the hypothesis ("Early posting garners more votes") is likely to be false. The "p" value is a measure of the confidence in the correlation. Lower p's mean that there is less likely to be an error. Statisticians think that a p < 0.05 is a good value (if I recall my stats classes correctly).
Could it be as simple as "early posts are near the top of the page and therefore everybody sees them"?
Yeah, when you post late, some of the top pics already have 20 or more votes. That's a lot to catch up to. Some people will return to the thread and vote for new entries but a whole lot won't.
I'm just hoping greenlight time will get moved forward far enough so I can post before leaving for work; I have no net access at work and with the commute included, it adds a good 10 hours to the time before I can post if, like yesterday, the thread hits the main page just after I've gone out! |
|
|
| |
|
|
|
|
|
|
|
Post subject:
Posted: 04.01.2006 03:12
|
|
Joined: 02.02.2006 Posts: 136
Location: Waldorf, MD
|
|
| I just can't believe that some yahoos actually snuck in B&W pics in a color-themed contest. Sheesh! |
|
|
| |
|
|
|
|
|
|
|
Post subject:
Posted: 04.02.2006 10:06
|
|
Moderator
Joined: 06.01.2006 Posts: 1600
|
|
i've entered both early and late and while the earlier do appear to have an advantage concept and execution count for more here.
what i really don't understand about the voting is when my on theme entry gets pounded by an off theme crappy lame snapshot. or when my barely on theme entry does better than someonelses right on theme one.
so in summary, enter what you want, when you can and just enjoy having the outlet. and tell your friends.i send email to people who wouldn't otherwise see the contests. |
|
|
| |
|
|
|
|
|
|
|
Post subject:
Posted: 04.03.2006 00:04
|
|
Joined: 02.02.2006 Posts: 136
Location: Waldorf, MD
|
|
| I hear ya Curious. Sometimes when I'm doing my voting I'm thinking "which ones capture the theme best" and sometimes I'm thinking "which pics have the best concept/execution"... depends on my mood at the time. I don't mind a cell phone camera shot that gets the theme right as long as it's not too chunky, but I won't vote for a fantastic photo where I can't figure out how it relates to the theme in question. And then there's times where you have a perfect storm of a great picture that also nails the theme like the winner of the "Green" contest. |
|
|
| |
|
|
|
|
|
|
