NewsWrangler's dashboard

TruthWrangler replied on February 25, 2008 19:34 to the idea "Suggest "My Predictions" Be Listed Showing "Latest" First" in Hubdub:

Thank you!

TruthWrangler reported a problem in Hubdub on February 25, 2008 19:28:

Question Creation Date & Suspend Date Bug

Several times I have seen a question with a "Created Date" that is later than either the "Suspend Date" or the "Settlement Date".

Here's an example where the question was "created" after it was "suspended":

Action history:
Suspended Wed 6th Feb 11:59pm PST : Suspend date reached
Created Today 7:28am PST

<http://www.hubdub.com/e/Market/Will_t...>

This apparently results in a question being suspended, with zero H$ (no 12-hour delay as well).

TruthWrangler marked one of Destry's replies in Hubdub as useful. Destry replied to the problem "Question was edited which changed conditions of question.".

TruthWrangler replied on February 24, 2008 19:22 to the question "Self-referential questions, why aren't they allowed?" in Hubdub:

Skipper -- on second third fourth thought, I think you're right about your question: "Will the 'no' percentage be higher than the 'yes' percentage at suspend date' easy to game?"

It can't be successfully gamed, and I think my answer goes way beyond the constraints of the Hubdub board.

TruthWrangler replied on February 23, 2008 22:16 to the question "Self-referential questions, why aren't they allowed?" in Hubdub:

Secondly, how can you have 55% and 49%... they equal 104%...

"I knew that ... I was testing you ... "

NOT!

Well ... I missed that one!

TruthWrangler replied on February 23, 2008 21:54 to the question "Self-referential questions, why aren't they allowed?" in Hubdub:

Ummm ... sorry -- I see that I changed the order of the specific question you asked, which was:

Will the 'no' percentage be higher than the 'yes' percentage at suspend date'?

Using the same example information as before, suppose that, at a minute before the suspend date and time, the percentage for NO is 55% and the percentage for YES is 49%. The two answers are "Yes" and "No".

At this moment in time, the way the question is worded, "No" is the percentage-winner (55-49%), and "Yes" is/will be the "winning" prediction that pays H$.

With a minute to go, and the same H$ 5,000 of activity in the market, I place a H$ 100,000 prediction on "YES". Bang! I win -- the result is that the "No" percentage is not higher than the "Yes" percentage at the suspend date. (In doing this, BTW, my thinking is that "YES" will win the percentage gain, with the result that "NO" will win me the big H$.)

Because the person settling the question would have to go through all of that, and because it would likely cause a good deal of user confusion, and because anyone with enough H$ could control the outcome of the question, questions like this (which are very, very clever!) aren't allowed.

(A less-clever question would have a "negative" in it -- like "Will Barack Obama not be elected as President? Politicians sometimes word legislative bills in that manner, so that the legislator has to vote "no" in order for the bill to "pass" -- then the politician has to try to "explain to the public" why s/he voted "no" on a bill they were in favor of!)

TruthWrangler replied on February 23, 2008 21:11 to the question "Self-referential questions, why aren't they allowed?" in Hubdub:

Maybe changing the words might help in explaining it ... if the question had two answers, but instead of "yes" and "no", the terms were "Apple" and "Orange", the question would be:

"Will the 'Apples' percentage be higher than the 'Oranges' percentage at the suspend date?" (And the two answer choices, instead of "Yes" and "No", were "Apples" and "Oranges".)

Now, suppose at a minute before the suspend date and time, the percentage for Apples is 55% and the percentage for Oranges is 49%. At this moment in time "Apples" is/will be the "winner".

With a minute to go, and say H$ 5,000 of activity in the market, I place a H$ 100,000 prediction on "Oranges".

Bang! "Oranges" wins, and I get an amazing return! -- (assuming no other predictions are made after mine ;-) ... and I have successfully gamed the system.

The same would hold true with the "Yes" / "No" choices:

"Will the 'Yes' percentage be higher than the 'No' percentage at suspend date?" -- and, at a minute before the suspend date and time, the percentage for YES is 55% and the percentage for NO is 49%. At this moment in time "Yes" is/will be the "winner".

With a minute to go, and the same H$ 5,000 of activity in the market, I place a H$ 100,000 prediction on "NO". Bang! I win. (In doing this, BTW, my thinking is that "NO" will win.)

I could also have placed the same amount of H$ on "Yes" -- and, in this example, I would also win -- just a little bit less ;-)

I win because I controlled the outcome of the question.

TruthWrangler replied on February 23, 2008 19:25 to the question "Self-referential questions, why aren't they allowed?" in Hubdub:

Skipper -- The question may be able to be settled, with, as you said, one percentage higher than the other -- however I think the first part of Nigel's answer is the applicable portion.

He said, "Self-referential questions are generally either very easy to game or impossible to settle."

In this instance, that the question is so easy to game is why it's disallowed.

TruthWrangler replied on February 23, 2008 18:04 to the question "Self-referential questions, why aren't they allowed?" in Hubdub:

== I'm sorry I had to explain this to you twice 8^) ==

Asking questions more than once can be a 'good thing' ...
(except, maybe, the proverbial 5-year-old repeating "why" to *everything*)

TruthWrangler replied on February 23, 2008 00:03 to the question "How can we improve My Hubdub?" in Hubdub:

Duh! I still haven't found the calculator!

TruthWrangler replied on February 22, 2008 23:56 to the problem "Suspend Date/Time Bug" in Hubdub:

Just wondering ... can this forum post topic be edited to "Suspend Date/Time Bug" ?

TruthWrangler replied on February 21, 2008 22:56 to the problem "Market with negative H$ balance (bug?)" in Hubdub:

Yes, (thanks!) -- I did mean "activity".

I only commented on it in case it was a (potential) beta-bug -- I had not seen anything like it before.

TruthWrangler started following the idea "International spoiler-alert" in Hubdub.

TruthWrangler reported a problem in Hubdub on February 21, 2008 22:26:

Market with negative H$ balance (bug?)

How are markets with negative balances settled? Or is this a glitch?

This market as a hd$-2,144 balance:

<http://www.hubdub.com/e/Market/Who_wi...>

TruthWrangler replied on February 21, 2008 03:09 to the problem "Suspend Date/Time Bug" in Hubdub:

You're right -- I ought to have said "suspend date".

Thanks for the catch!

(also agree on the s-l-o-w submission time)

TruthWrangler replied on February 21, 2008 02:48 to the question "How can we improve My Hubdub?" in Hubdub:

Maybe the refresh is a browser issue. I'm using Firefox 2.0.0.12, with Vista (but you guys know that, right? ;-)

TruthWrangler replied on February 21, 2008 02:41 to the idea "Display Original "Likelihood Percentages" on Question Page" in Hubdub:

Infernal Machine, I agree that unequal likelihoods are, or can be, problematic ... and I sometimes wonder "why" a percentage is set the way it is.

OTOH, sometimes I can only go with a "gut feel" ... I may not know what the exact odds are -- but I do "know" they aren't 50-50. If I had to try to explain why I set the odds the way I did, they would probably all look like this (but with varying likelihood ranges:
<http://www.hubdub.com/e/Market/Will_H...>

... in short, I wouldn't be adding information of any real value.

Maybe there will be other suggestions that help with this.

TruthWrangler reported a problem in Hubdub on February 21, 2008 02:31:

Suspend Date/Time Bug

When creating new questions, setting settlement dates don't always "take" when the question is finally published.

This seems to be an intermittent problem -- and may explain why some questions that *need* settlement dates don't have them ... the author may have set one, but it didn't take.

One example was:
<http://www.hubdub.com/e/Market/Will_a...>

In this instance I sent an email to Hubdub, and Rob Jones took care of it (and noted it might be a bug).

I just noticed the same problem in a new question by another user:
<http://www.hubdub.com/e/Market/Will_t...>
(see comments).

TruthWrangler replied on February 21, 2008 00:52 to the question "How can we improve My Hubdub?" in Hubdub:

I really like the new pull-down menu option for comments ...

I'd like to be able to use the browser refresh on all of the pull-down options to have the most recent selections available.

As it is, one has to do a convoluted click-on-something-else-then-click-back-to-what-you-want in order to accomplish a refresh.

TruthWrangler replied on February 21, 2008 00:25 to the idea "Display Original "Likelihood Percentages" on Question Page" in Hubdub:

I didn't think of that Average American ... that would likely be the BEST reason for having this feature!