Discussion in 'General Discussion' started by dndgeek, Jan 31, 2016.
What are the criteria to get a Premium Badge?
First come first serve. Cost more. They post dates of available for later this year already.
Re-reading the thread again I think I'm mostly in agreement with what @dndgeek wrote.
Have a lottery system for each access level (not badge level, since gold and silver are really the same access level, just different loot levels).
You could have the lottery say on the 1st Saturday in February. All badges purchased prior to say 1pm of the lottery draw is in the draw. Have time slots for each draw (of say 10 or so people), with the slots being 15 minutes log to submit your event registration request. You could process 40 people an hour that way. Have this go on for like 14 hours a day until the initial draw is completed. Then it's open season after that.
I think this is simple enough for folks to understand and simple enough to not need a huge spin up of AWS. No complex conflict resolution/queueing algorithms. Just trying to beat a smaller herd of folks to the events you want. You can still load up your cart with events and then reconfigure as things fill up.
Then you can focus effort on a good filtering/search system for the event catalog so people can easily find events if/when the ones they want go away.
But I'm a simple man that can't code for shit so my code tends to be simple so I can understand wtf is going on with it.
For what it's worth, I actually think the ability to load conflicting or overlapping game slots into one's cart is a good thing. It allows partners or groups to use one or a few badges to preregister, allowing some people to buy regular at-the-door passes while still being guaranteed one or a couple of events. Given the impossibility of predicting what one might actually succeed in acquiring, it's also a customer service to allow the chance to hedge one's bets a little. So I found that to be one of the positive differences this year.
@gizmomathboy :just to clarify, I was saying that no registrants should need to click on anything on the day of event registration. A program could just process all of the requests in everyone's carts at the cut-off time in minutes, allowing results to be displayed quickly. That way everyone has ample time to load their requests (a week or two?) and not be required to be present at registration time, which alleviates a lot of issues, such as availability, connection speeds and problems, etc.
Another thought would be to add a second tier in the gold/silver level and have an early bird badge sale, giving the diehards and those paying attention better placement within the gold/silver level.
That would work as well I think. Something more determinate in terms of your "place in line". There's still luck to it, but don't need an itchy trigger finger.
1000 Event Catalog Finalized Day: All events are in and published for people to start perusing. Count = 1
2000 Ranked Registration (re)Opens: If Count >=3, Goto 9000; Otherwise, allow users to start putting events in carts. When adding, users must rank them according to preference.
3000 Cutoff Day 1: By this day/time, events are submitted by users, ranked by preference. Ranked Registration Closes.
4000 Cutoff Day 1 + 1 hour: The computer uses algorithms to assign tickets. It does this by lottery. The lottery changes each pass - that is, it does not process a single persons list first. It processes everyone's first preference in a given order. Next pass the order will change. If a person gets shut out of a game, that individuals chances in the next lottery round increase (thereby increasing the chance that they will get their next preference).
5000 Cutoff Day 1 + 1 Day (assuming the algorithms finish in 24 hours): Users can now see what they've received tickets for. Balances are adjusted.
6000 Users bitch about tickets received.
7000 Increase Count by 1.
8000 Goto 2000.
9000 Open Free-for-all Registration.
I can deal with this one, too. The magic in 4000 could be a bit dicey but not *too* complex. The complexity really being the handling of the folks that got their first "available" preference in the first pass. There's room for some haggling but another system that is "better" than 1,000 folks mashing enter at the same time.
IIRC GenCon has multiple tiers for event registration and housing.
Instead of a lottery what are your thoughts of no pre-registration and no event catalog prior to the con. Show up and people run and play what they want.
Couple this with a single price badge for the con.
I have heard all sorts of things like they are looking to use GenCon's software, or X con's software. My mind is all fuzzy with the whirlwind discussion.
While I am not a big fan of higher badges get preferential treatment, I do understand it is a business and the revenue from those high priced badges help keep the rest of the badges less expensive. Or at least in theory.
I have thought long and hard about how this could be improved to be one of the best cons ever... at least at the registration step.
So I am going to re-iterate some of what was said before and add some steps to it.
One person per account. Either allow group linking via a friend setup OR allow purchasing for sub-accounts on the main account.
Once the event catalog is release, people can add tickets to their cart.
They can add tickets for sub-accounts so you can do group allocations if you are going with your child or a friend.
If getting for more than one account, add a check box to say only pick up if enough tickets exist to satisfy the # of people
They would order events in the order they are looking for by priority.
Event registration would close 30 minute's before the system would attempt to start processing request. The published run time would be an hour (or two) before the cart system would open back up for cart processing. This is so people can plan on when to be online to pick up additional events.
All carts that are in the system, will get assigned a random number between 1 and 1000. Yes, there can be duplicates but it does not really matter it is for rough and random sorting round to round.
All platinum badges would be assigned a starting loop of 1 (this gives priority for those badges for their #1 event).
All other badges would be assigned a starting loop of 2 (starting loop number can be increased to provide higher badges more events before lower badges)
On subsequent loops, all badges would participate.
While more accounts in this loop.
while (assigned at least one event and event remaining)
check to see if it available and it does not conflict with a previous event
Yes - Assign to account and remove from cart, break from sub-loop
No - Leave in cart and mark priority as processed
update account loop number by 1 (so this account will not process until next loop)
update random order. This randomized the order for the account for the next loop
Repeat process until a accounts with saved carts are processed.
Email all notification to all individuals
Post processing phase:
Enable regular cart system so people can fill in what they missed.
What that really means, is the platinum badges will have the best shot at getting their 1st or 2nd choice (one loop through - could be more it is up to GaryCon admin)
Then everyone else will randomly have access based on the random number generator to get an event. It will be their top event that is still available and they would get one event per loop. It could be their 1st, 2nd, 3rd... does not matter.
It allows you to schedule backup events for certain time slots if you miss out on a higher choice.
It also should eliminate the server overload issues making this a background process. People will not be clicking and refreshing continually.
In retrospect, was there an opportunity missed this year to do a simulated load test using what had been loaded into carts prior to opening up registration? I realize that would not have provided enough time to fix all the issues that such a test would have revealed, nor would it have addressed things like incorrect coins on accounts and such; but it would likely have exposed the need to delay the registration to analyze the load constraints and resolve in a manner that did not result in this calamity. All that this would require is a QA copy of the environment connected to the e-commerce service and some code to submit x% of total carts within a narrow time-frame. I suspect that this would have exposed the weak link in the system and prompted an upgrade to the e-commerce server, which seems to have been the culprit.
Hindsight is 20:20, I know, but those who do not learn from history are destined to wallow in failure. Regardless of what system GaryCon implements next year, if it is not load tested, it very well could fail just as catastrophically. Or worse.
I really like the idea of an automated registration system that just randomly processes the cart contents instead of having people scramble to beat each other at check out. It would eliminate the complaints about people not being able to be at a computer with a good internet connection when registration goes live and certainly wipe out any "fairness" complaints. In previous years, I have been remarkably successful at exploiting the registration system, but I would have no complaints about going to a truly egalitarian system like Jim Skach is proposing. I have no idea whether that is a hard system to code, but it sounds great to me.
Personally, I think doing away with the "pre-loaded cart" is the best way to go. Let everyone register for the events they want by adding the items to their cart when registration opens. This way, there aren't 1000 people all clicking the same process at the same time, doing away with the pre-loaded cart forces people to spend time loading items into their carts before processing. People who only register for one or two games are going to load and process a bit faster, but won't be getting a lot of events. Someone who is trying for 7-9 events would take longer so processing is a bit more scattered. You could still let platinum and diamond members pre-register their events before the registration even goes live. IMO this is the easiest way to do it.
My 2 cents...
I like the over-all concept of how it works now. The only changes I'd like to see:
5 tiers of badges, Diamond, Platinum, Gold, Silver, and Copper. Only 1 badge per account.
Each tier gets 2 days access (not hours) where Diamond opens Feb-1, Platinum opens Feb-3, Gold opens Feb-5, Silver opens Feb-7, and Copper Feb-9.
# of badges are restricted to 20 diamond, 50 Platinum, 200 Gold, Unlimited Silver and Copper.
Prices similar to this year for Diamond and Platinum would allow the Copper level to be much cheaper than this year's silver for people that just want to get in pick up games or can only attend 1 day but want to register for a game or two.
I HATE the pay per ticket thing for many reasons. It makes the GMs act as cashiers. Having to buy generic tickets. Games being delayed because someone had to run and buy a ticket. Getting bumped from a game because you lost your specific ticket.
I can tell you from experience that I can win in that system. I've gotten into an unholy amount of desired games in previous years, including a lot of VIP events. An automated system working off preloaded carts would take my sinister abilities out of the equation.
This is the first year pre-loaded carts was an option, and the last two years, the sight crashed as soon as registration went live. Like Buttmonkey indicated, that system was exploitable by those savvy enough to do so, and I admit I took advantage of it as well. The concept of pre-loaded carts and a hands-off submission for the initial event registration seems as fair as any process and would also afford an opportunity to load test the system prior to actual registration.
to pull back the curtain, at least on the method I used, it is simple to open a new browser window for each event you want, then refresh one of your browser sessions until you see the add to cart box. Then it is a matter of clicking on the add to cart, tab to next window, add to cart, tab, add, tab, add... check out. And you are done before the system crashes on everyone else, and with probably 100% success rate.
To be perfectly honest - HATE it
It's a simpler registration (ie, none), and cuts down on registration complaints.
But it's either a nightmare to promote/signup events onsite or people just wind up settling for the first thing they stumble onto that sounds at all appealing, because there's too much to sift through (probably both). Lots more 'dead time' and much harder to coordinate with friends, because you can't reliably know start/end times.
That's more sophisticated that my methods. I'm impressed.
Separate names with a comma.