I'll agree that a 503 error code is annoying. At noon, when I pressed "submit", I had a "waiting for response" for over 5 minutes and decided to refresh, in which I was asked to submit again. It threw me in at 2500, and the person beside me who had to do the same thing, was given a number of 4000. The three other lists submitted in the same house hold were earlier, between 400-1000. Out of 5 people, the results were varied, as we expected. We all knew there was a risk that some of us were going to get a high number, in fact probably most of us when you are competing with thousands at the same time. Even if the system was automatic, or didn't have any errors or lag, there is still a chance I would have ended up in the 2000s, or even higher. Except for one thing, my friend in the 4000 got everything on her list. I got most of what I wanted and most things I couldn't get was due to conflicts. In fact, I am already thinking about returning some tickets because I think I might be too overbooked, if I want to enjoy the other aspects of the Con that don't require early registration.
I guess I am wondering, if people are so defeated by what they didn't get at event registration, why is attendance so high and why is it growing so much each year? If everyone not getting exactly what they wanted every year is ruining Gen Con, then why aren't people dropping off in droves?
I think one of the keys is designing a good wishlist. You are allowed 50 slots, USE ALL 50 SLOTS!!! I was on another board in which someone was ranting because they got none of the events on their wishlist - They put 4 events in their wishlist.
I usually divide my day into 5 hour chucks, morning, afternoon and evening. I find 5-7 events for the morning and afternoon time slots for each day. After wishlist processing I go back and fill in any holes and find evening events I am interested in.
Lots of people come for just one day, and walk around the dealer hall.
Lots of people just play Magic, come to do the Costume Contest, the Anime films, True Dungeon, or enter into some long tournament that doesn't typically sell out instantly.
I mean - around 6,000 is the highest number I've seen quoted for the waitlist. There are many, many more people than 6,000 coming to GenCon this year.
Dumped 42 events into my wishlist, got plenty, dropped the ones I didn't need.
Of course, this is sort of an obnoxious solution - as if everyone does this then it keeps others from getting into things that will actually come available, when they are dropped 1-2 hours in.
I do the same thing, but make sure most, if not all, of my backups overlap one of my "first choices". That way, it's impossible to get more than 15 or so events on the first pass, when I'm looking for 11 events total.
If it wasn't for the 50-event limit, I'd probably 60-70 events on my Wish List, since I'd have *every* True Dungeon run of the same type that overlaps the first one on my list. As it is, I'm only able to put about 8 True Dungeons in the same slot, and got shut out of one of them in the initial rush.
I got "errored out" and ended up with 6239. I got a lot of what was in my wishlist, though, including some True Dungeon and Helix to Helix. I'm still voting / hoping for a Waiting Room kind of setup like SDCC does for badges, where you log in up to 2 hours before the opening bell, and at Noon anyone "in the room" gets randomly assigned a slot, and any attendees who log in after that are placed as they come in. Let's face it, it's random now, just not in an orderly and designed way...
Dbachman, care to show your work? Without opportunity for review, it doesn't mean much. =/
Does it mean as little as your response? I outlined the algorithm in the thread already. What are you asking for...the tables, the code...what? What exactly is your background such that you can offer code review and yet not see how simple the parameters are for modern computers to deal with? Fact of the matter is I did not even use professional level software (what I have at work) to test out the approach. If I had the results would have taken less time. The question here is your basis for questioning the results I put forth. Are you saying they are too good to be believed? That I didn't write and run code against a database of events, attendees, and wishlists?
All you said was that you created a database and "ran" it. You didn't say what you were going for. You didn't show any process. You didn't give results other than how long it took you to run. Is that what you were going for? Your prior comments and closing opinion indicate that you were trying to find a more equitable process, so it doesn't seem like it.
Just saying that you ran some tests doesn't tell us anything because we don't know what you tested, how you tested it, what the results were, or how you came to your conclusion.
My group is out on the east coast. Even optimizing our submission process as best we could, the best position that any of us got was just under 3,000.
While I cannot speak for GenCon's systems, it would be difficult and flawed to set up a system to compensate for geography for a few reasons:
1. When working over the internet, things are rarely as direct as you may expect them. For example, a traceroute of traffic going from my current computer in Ohio to gencon.com's servers (reported in Ottawa) has it going through routers registered for Cleveland, Washington, New York, Colorado, Philidelphia, and British Columbia before arrival.
2. How would you know how much to compensate? There is a very fine line between ineffective and wrecking someone.
3. Adding additional lookups and calculations would only slow the servers that are already failing to accept so many requests as it is.
I will have fun at gencon, regardless. I have some sense of which events are worth showing up with generics (anything that starts before 10am=very good chance someone slept in). There are so many great events that I could sit at a random table based on availability and play something good. I can just walk around talking to strangers and taking pictures. It will be fun. And we can't all participate in the same 6 player game at the same time. No way around that.
In this day and age, 503 errors and a system that takes 2 hours to process 6000 submissions just doesn't make sense. AWS provides explicitly this service. One of their case studies is a university system that publishes test results, it gets hit with a massive surge, once a year, at a designated time. It is a very strong parallel to this situation. I can't comprehend why anyone would drive a business as large as gencon, with a homebrew system. With a working technical system in place, we could discuss a variety of approaches to the actual distribution of the contested selections, but without a working system for handling the transactions, randomly assigning "you get everything you ask for" and "you get nothing" makes as much sense as anything, and we are all left to game that system as best we can.
Geographical location may not make a huge difference anecdotally, but it's hard to argue that distance has no effect on latency.
It's also worth pointing out that I don't believe that implementing a process like this would solve any of the existing issues or necessarily make anything more "fair". I'm using it more as a point of discussion.
Compensating for geographic location is a fine tuning detail. It doesn't make sense to optimize at that level of detail when you have issues that are orders of magnitude larger.
Geo location is insanity. That's like saying the most 'unfair' thing is related to the speed at which you can click. That alone is nuts.
I see two issues: 1) Why process the entire wish list based on order - instead process top event for everyone in the queue...cycle back through the queue...until everyone who clicked from 12:00 through 12:03 is done.
2) Create a way to sign up for True Dungeon where you pick a range of times instead of one particular event.
I live by Cleveland, Ohio, which is 400 miles away from GenCon's webserver. But somebody in Delta, British Columbia, which is 2200 miles away, with Peer 1 as their carrier will have a notably faster connection to Gen Con's server. Geographic location is just not a usable variable.
I didn't read this in these responses but does anyone know why gen con doesn't just use the wishlists before registration opens. That seems like the purpose of the wishlists. Its a lot less nuisance and you can run a priority program on it so every ones 1st choice gets satisfied before every ones second etc. down the line. You can even chain events and people so that if one can't get in all don't or all do depending on the outcome requested. Tell people they have three weeks to submit wish lists with automatic reminders every few days, you can even build the wishlists into the system automatically so most of it is complete before registration even begins.
I think my biggest complaint about the current system is that people in rural communities automatically have slower responses via their ISP's due to network latency. A newer computer automatically has a faster response to a mouse click, in general, especially more expensive computers. Combine those two issues and it essentially created a system that favors rich people even though it wasn't their intention.
And if you are living on a reservation forget about it. latency rates on reservations are notoriously bad.
I know how thoughtful gencon is and how important they prefer treatment of those who may be considered outsiders in communities, I'm somewhat surprised that the latency issues in poorer communities weren't considered when they determined how to treat those requests. i thought thats precisely the reason they moved to random assignments in the hotel queue
Again, my original question was asked out of simple curiosity and was built on to form a discussion, not as a viable solution. As another user mentioned above, people in some areas will just have a lower latency than others, with various factors being the cause. I was more interested in determining whether or not there were any sorts of balances made with any of this in mind.