InfernalMachine's dashboard

I agree. After all, there's an All tab to look at every question (in some sort order), so there's no reason to tag something as X and General. But a question like number of viewers of Super Bowl could legitimately belong to both sports and entertainment. General should be reserved for questions that don't fit any of the other categories, and shouldn't be paired with anything else.

comments part 2

Settlement: Good you make clear about the delays in settlement related to editors having to sleep.

In paragraphs 2 and 3, the term "suspend time" is misleading. You mean the time when the question is determined by real-life events.

Hopefully in the future you will have editor assistants who live in different time zones who can adjudicate while Midlothian sleeps.

"Number of times" questions should be discouraged. If the settlement source is public news sites, there's no guarantee they will count that word or phrase. You have to then go to public transcripts or watch the video, neither of which may be available. (ie things on you tube usually get yanked for copyright). Anyway, should an editor have to slog through pages of transcripts (I guess you could just run a word count) or minutes/hours of audio or video to settle an event like this?

Disputes: Is it possible to add a new Category to Get Satisfaction to collect these dispute posts. As use of the forum grows, it's going to get harder and harder to find things and keep up.

Voiding questions: Could Hubdub duplicate this forum's approach to topic creation? That is, force a search on the topic before you can opt to create a new one - would help with duplicates.

Still think that question creator shouldn't set likelihood values (unless they are barred from betting on question). When question first goes live have ability to vote on option exactly once with no payout. When 50 or 100 have voted, use these as initial likelihood, and open betting. User could keep track on their page of which questions they've voted on are now open to bet on. (To be clear, this is separate from the new mechanism for vetting new questions. It follows it. And users are voting on one individual option like yes or no within the question.)

Why not require conditional questions to have an option that, in the example given, Edwards does not drop out.

Change of Circumstances: "Any official changes to the outcome of a question after that question has been settled will be disregarded." As stated, this contradicts the ability to appeal the result of perceived settlement error. This is a tricky issue which requires extra thought. You want appeals for incorrect or too hasty settlements, but you don't want this to go on forever. Perhaps limit the time for filing a settlement appeal.

Creating Questions: Personally, I feel 12 hours is way too long to freeze out creators. If cheating is so worrisome, bar the creator from betting on own question.

As in part 1 of my comments here, I want lots more guidelines and rules for question creation. Carelessness at the beginning causes much grief and trouble later. If the question creation process is properly explained and guided, a huge burden on the admins and on the predictors will be removed.

I'd suggest that a worthy project would be a redesign of the question creation process and page(s). Make it much more guided. Come up with types of questions and have appropriate forms for each (like the current differentiation between yes/no questions and multiple choice). Reflect the result of these in the page that the predictor later sees.

TV Shows: Not sure I like the voiding of questions where a tie which isn't anticipated occurs. These contests are like sports races where ties are very rare. (Many games by contrast have ties as expected potential result, and questions that reference these should/must have a tie option or be considered invalid.) Is there no mechanism you could build into the program to account for this and allow an adjusted payout for the 2+ options that did "win", versus all the options that lost? I'd be happy to work out the details for you. It wouldn't be the "if right, you win" amount, but it would still be preferable in most cases to the wager amount.

Weather: How about all weather questions with temperatures must state both Celsius and Fahrenheit (link to web temperature converter on question page). Similarly, both inches and centimetres for rain/snow. Both MPH and kph for wind speeds. And so on. If I can't look these up when making a question, why am I bothering anyway.

Wind chill factor not a temperature. Agreed.

Sports: Would you consider an idea I put forward elsewhere in this forum that in-game questions be the only questions that can be duplicates? That is, if a normal question exists about an event which suspends at start of game, let it stand as such (because people who bet in its market did so assuming it was not in-game). Create an in-game duplicate (with NO 12 hour freeze on cash-in -- and since an admin created it and will judge it, they can't bet on it), maybe stick it into a new In-Game category so people can find it, borrow the suspend time odds from the question it duplicates, and open it say one minute after the normal question suspends. OTOH, In-Game questions that are original can be set up to the admin's pleasure, provided they are labelled as In-Game, and normal questions which later duplicate these can be rejected or voided.

Members of the school of thought that In-Game is "cheating" have their points. Perhaps a purists' Predictors Club leaderboard for those who have never bet in-game could allow for peaceful co-existence for all.

Politics: As we've seen with Super Tuesday, delegate counts are not as simple as they sound. Some states take a long time to come up with these figures after the primary/caucus. Perhaps the onus is on the question creator to actually research the event in question, explain it briefly with their question, and discover and specify an actual likely settlement date, or else the question should be voided ASAP (before people have got too involved in betting on it). Personally, I think primary questions should ONLY refer to popular vote, because these are usually able to be settled within 12-24 hours. The Super Tuesday total delegates question for each party should be banned in future years because of the complexity of settling it. The November Presidential election will have similar problems if it's anything like 2000 (or even 2004), so people should be made aware that it may not be settled until January of the next year.

Business: Agree with the individual stock ban in general, but will miss Google/Yahoo type questions, as they are newsworthy beyond just the financial sector.

About major indexes: What are the major indexes? Specify them. Allow users to push for new additions (it all comes down to how many people will participate). Any way to automate the creation of these questions for say the DOW daily (open 2 days in advance?) by the business editor?

I just realized that the index questions are very similar to the Sports games questions (ie they can be seen as games with final outcomes which fluctuate in potential all through their duration). Should, for example, DOW questions close when the NYSE opens, or say 1 or 2 hours before it closes, unless they are run In-Game by a Hubdub admin? If not, it seems Sports questions are held to a much higher standard than Business:Index ones are.

Overall: Very helpful guide, and the rules seem appropriate.

I shut up now.

Comments (part 1)

It's very specific, seems fair to me, and overall is excellent. Should be a great step forward.

One major quibble about the rules. It still doesn't place enough onus on the question creator to really think about the question they are asking (have I considered all reasonable options, have I made it fair and straightforward to settle, will settlement drag on for weeks after the suspend date,etc.)

Final draft should have a summary of most important points at beginning lest lazy readers not get past first 20 lines.

Outages: perfect. Act of God. The risk we all take.

Time Zones: Still too confusing. Stick with one (PST). Add a widget to the question creation page that calculates time zone changes (if not a widget, then a link to 1 of many TZ calculators on Web), so the PST date/time for event and suspend date/time are entered correctly. Why not have a field to enter the question locale's Time Zone (major city or standard abbreviation of time zone - we could have a list of them in the faq), which would appear next to the additional text on the prediction page. Also the suspend time PST should be up there too. If the above widget is available here too, the predictor can figure out how PST relates to both event time and their own time easily.

Suspension: "While a question is suspended, it will not be possible to place any new predictions or to cash-in any existing predictions on that question." Does this mean that the cash-in loophole is going away? Good.

Great that administrators (this includes editors?) can change things, but should also state that it is the creator's responsibility to get things right first time or have question sent back to them.

"The Suspend date and time should not be used as a guide as to when the question will be settled. E.g. If the question is ‘Who will Edwards endorse?” and the suspend date is February 5th then the question should not be interpreted as “Who will Edwards endorse by February 5th?”"

I don't get this. It says what isn't the case, but leaves open to doubt what is. Suspending a question (by the creator) potentially months before a settlement is bad question-making practice (especially if cash-out is suspended too). The question should be reframed by the creator, not endured by the predictor. In other words, I think these kinds of questions need to have an explicit decision date (as opposed to settlement, which is a Hubdub mechanism) built into the question. "Will X happen?" is a bad question, made worse by having an arbitrary suspend date. It should be sent back. The proper form for these questions is "Will X happen by Y?", and the suspend date Z should be close to date Y, unless the creator explicitly explains in the extra text that your money will be potentially tied up from Z until Y and also explains why the question is framed this way.

Types of question: Basically sound. Personally though, I will miss Hubdub questions. I feel our stats are a form of news source, and they're public to members of the community. The self-referential questions, the ones with strategies available to ensure one outcome over all others (ie $ in trade), and questions that allow one or more members to both predict and determine the outcome are not fair and should go, but for the reasons I've stated, not because they're Hubdub questions. I think Hubdub questions build community. Perhaps you might consider easing these restrictions at a later date.

(to be continued)

me too. we've beaten the poor server into submission.

Good point! Well, maybe scrap the fixed regions and just make sure the closing time (if fixed) is in open hours. But it would work as is, like questions that suspend before the result occurs, like cutting off bets an hour before the stock exchange closes. Each question could have its own hours of inactivity (although it begins to get complicated - think of all the different times the creator has to enter and the bettor should check for before wagering money).

Anyway, I'm not sure that 24 hours are unfair. I'm just throwing the question out for discussion.

New York Times (but NOT the AP figures, which are an estimate -- they even add in "guesses" of how superdelegates will vote). But their delegate page, which is very careful to only enter official values into their tally. I should add that some of the state parties don't decide until later in the season how all of their delegates are divided up (not sure if they affect the questions in question). CNN figures include all past contests and declared superdelegates. MSNBC includes previous primaries. The best site I have seen is called "2008 Democratic Convention Watch" (not associated with the Dem. party though). They show the allocated delegates for each state for the 2 candidates, and the number to be still allocated. Currently a tie with 300+ still to be allocated for Super Tuesday. But it's a blog, not an official news site, though it's worth a look.

Well, I'll try.

1. I'm not saying because you live in such and such a place you can't vote on particular questions. I'm not even saying that you can't bet or request a cash out at any time, just that your request wouldn't be processed until that question re-opened, just the same as night traders do with real stock markets. But you'd be only betting on the closing figures, with no sense of what the "overnight" action's effect on opening "prices" would be. You would also be able to cancel a pending order at any time UNTIL the market re-opened.

2. Again, you would not be shut out of the Dow questions at all. But the time window when Dow questions were "active" would be less than 24 hours, would contain the Dow's hours (say 5am EST to 9pm EST, but closing totally at 4pm on the day in question), would allow night orders for the morning (ie bets and cash ins between 9pm and 5am EST would be recorded (and able to be cancelled until 5am) but not processed until 5am). If you live in Tokyo, this might be inconvenient, but less so than actually running a NYSE portfolio from there (ie our market is open longer hours than the NYSE).

The regional aspect is just that the events in a region (sports/stock market/political announcements/etc) tend to happen in local daytime, so if these group of questions "sleep" for awhile it should be when people in that region usually sleep. If you live in a particular region you can bet on any questions in all the regions, but those questions might be dormant for a portion of the time you are physically awake. There would always be questions open in at least two regions (of 3 or 4) at a time. The other idea about regions is that not only am I more likely to be awake when the important things happen locally (not me personally - I live in Toronto Canada but inhabit Tokyo hours) but I may have more interest and knowledge about these questions. For example, I don't vote on German football games as a rule.

The only reason I came up with "regions" was to stagger the "sleep" times of questions, so that there's always questions open. The reason for suggesting such down time was to address the issue of whether or not having markets, which are volatile both in their speculative nature and in reacting to outside information, open when the people most involved in them can't react to changes affecting those markets is potentially a bad idea.

It's a philosophical question mostly, but there's also the fairness aspect. Say the question is Will J Lo have babies by Sunday, and being a night owl I happen to catch the breaking news at 4am EST Friday that she's on the way to hospital. Is it fair that I (and a smallish number of Europeans and Asians) then dump lots of cash on Yes, if the majority of people betting on this question won't find out for 3 more hours. I know the expected likelihood has gone from say 30% to 80%, so I get in early on Yes sending the odds up to 45%, wait until everyone else hears and then sends the odds up to 90% (briefly), and double my money by cashing out, even though I have no idea whether it will turn out to be a false alarm. A closed market until 9am would have me and all the 7am newspaper readers and breakfast television watchers and texters and blog readers all at a same level of knowledge when the market opens, all gaining the same percent because all the early bets are processed as a block. Otherwise I get huge gains and they wake up to a market that's hardly worth betting on (money-wise) except for the first few in.

Individuals (unlike companies that can run shifts or operate foreign bureaus) have to sleep and feed. They cannot play a market 24 hours a day for weeks on end. They require their local stock exchange to work with that, and shut down sometimes, because a market becomes too risky for an individual if you can lose half your wealth overnight while you are sleeping. (Corporations can stay up and on top of it, but markets close for the sake of the small investor, among other reasons.) So why don't ours, for the same reason?

Sorry. I forgot the other half of my argument. The other advantage of regional markets is that no editor can be settling or voiding questions 24 hours a day. People who get impatient about getting their winnings immediately forget that the editor may be half a globe away from them. Having regional editors would help with this.

I agree. Sort of. Doesn't this possibly making cheating by collusion too easy? I mean that could go on outside HubDub already, but this opens up new opportunities to form cartels easily. I'd be more comfortable with this if the chatting was public and archived.

Unfortunately, public chat has its own problems, with spam and rudeness, etc., and it would use up human resources to monitor and arbitrate that would perhaps be better used elsewhere.

Isn't this better left to the HubDub Forum at
http://www.hubdubforum.com/

(Am I not a good acolyte, TarHeelV?)

Patrick

Yes. And ways to sort (or filter) the various lists (settled/void/etc) according to topic or transaction time or profit or settled date.

Accuracy over speed. Yes it's nice to have your winnings promptly to bet on other questions, but mistakes in settlements can be very upsetting at the time, and lead to real alienation and ill will. The (virtual) contract you take when you make a bet is that the question will be adjudicated fairly and that you will be paid your winnings. I think that, compared to this, speed of settlement (within reason) is secondary. If you're concerned that there's enough free-to-bet capital available at all times, then give new users more cash, and consider putting limits on maximum bet or total wagers on one question.

Oh. The post is just above. Duh. Silly me.

They probably suspended one of the questions because it was a duplicate, That's mentioned in the faq. The idea is to not spread the liquidity to too many markets around the same essential question.

As for HubDub's attitude towards your playing style, I'd guess they encourage it. I had some discussions along these lines with some of the admins and they were very encouraging. The problem they're after is intentional cheating, as far as I know.

That's a great point you make about the corrective nature of short term gain-seeking on the larger market. All the individuals, each selfishly seeking their own gain, are unbeknownst to themselves chewing through the falsehood of appearance or hype or wishful thinking or fear to reveal the pattern or facts or truth underlying the confusing and contradictory impressions.

I agree that the cash out freeze is too long. Maybe a couple of hours might be more palatable. (I have a post elsewhere in this forum suggesting a way to separate the question creator from the setting of prediction odds which might make the freeze unnecessary.)