See tickets? That's crap, that is.

[j4]

Look, really, though, I reckon if enough of us joined forces we could write them a better booking system. I seem to remember thinking this last year, too (and this year's is worse).

We could make our millions, honestly (or at least have free gig tickets for the rest of our natural lives).

Who's in?

Comments

TICKETS

[(Anonymous)]

ANYONE MANAGE TOO GET ANY TICKETS YET.. OR A DECENT PHONE NUMBER TOO GET THROUGH ?

[mobbsy.livejournal.com]

It's a really scary problem that I've thought about in detail in the past. Most of the time you have nearly zero demand, then a few times per year you suddenly have thousands (tens of thousands?) of simultaneous connections, all trying to have a transactional card authentication and tickets allocation.

[juliet]

Hundreds of thousands.

The rough method they seem to use is:

- send vast majority of connections straight to "busy" server.
- allow small number of connections through.
- once you have a connection, your connection is prioritised (sent straight through to the real server) and you have 5 min to get through the 4 pages required (front page, which-ticket link, ID number, credit card). After that you're back to being one of the hoi polloi.

It's a reasonable theory.

What they appear not to have done is ensured that the "real" server can deal with the number of connections they're allowing through. So you occasionally get no-data pages returned, or "connection reset" or a brand new different "busy" page halfway through. Occasionally, this happens halfway through e.g. submitting your credit card details.

I suspect that the problem is somewhere in the connectivity part of things - that the "real" server doesn't have a sufficiently prioritised connection to the current privileged users, and/or to the credit card company.

In addition to this, to avoid what happened last year (people not getting a confirmation page & therefore rebooking & winding up with duplicate tickets), they've instituted a check on the ID numbers. If you've already been through once, you can't use those IDs again. Fair enough: except that if your transaction went tits-up halfway through, you have no idea whether your debit card did or didn't reach the server, & thus whether or not you've successfully booked.

It *may* be that for anyone whose IDs checked out & who got as far as the "paying" page, they've automatically allocated tickets & if the card failed for whatever reason they'll sort it out afterwards. One can but hope.

[j4.livejournal.com]

Most of the time you have nearly zero demand

Mmmrr, I think this is kind of irrelevant, surely you spend the rest of the year testing the system? Simulating the effect of gazillions of people hitting the system all at once?

Also, seetickets do manage other big things like, I dunno, Madonna at Wembley.

And Glasto has the pre-reg stuff, so you know the maximum number of people who can legitimately try to book.

then a few times per year you suddenly have thousands (tens of thousands?) of simultaneous connections, all trying to have a transactional card authentication and tickets allocation.

Last year there were 137,500 tickets on sale (for which 400,000 people pre-registered), and they sold out in 2 hours. I'm not saying it's an easy problem!

I am sure there must be ways of separating the stages of the process, though, so that e.g. at any point up to submitting credit card details you can still get bounced back to square one, but when you submit credit card details either it can't kick you out or if it does it makes it bloody clear whether your payment has/hasn't been accepted. I don't know how you'd do any of this, obviously, but I can't believe it's impossible. Not least because I think they actually did it slightly better last year. :-}

Of course the simple non-techy solution is to make it a lottery, all administered on paper.

[juliet]

Or even administered electronically. You have a month to register, as per currently; you can choose to register in groups of up to, say, 4 (to avoid the "I have a ticket but no one else I know does" problem).

After registration has closed, you pick the right number of tickets.

1-2 months beforehand, you do a second batch to allow for the folk who wind up cancelling.

Re last year: even allowing for the fact that they've time-limited the Magic Connection this time, the Magic Connection has been ballsing up. So yes, something has gone wrong.

[htfb.livejournal.com]

Gig tickets? Have you taken up "traditional boat" fixed-seat rowing? I don't think the bookings for the Great River Race open until next month, though, so it's a bit early to worry about your place.

[uon.livejournal.com]

I'm not sure the organizers really care much, given that they know the tickets will sell out super-quickly even with a crap booking system in place.

Apparently (http://www.nme.com/news/glastonbury/35595), Michael Eavis wanted to sell more tickets by phone this year, "to help young fans get tickets". Eh?

[juliet]

Yes, I was thinking this. They sell out whatever, so why care? Certainly why plough money (& it would be a *lot* of money) into a system that only operates for 2-3 hours every year?

(I think some form of "keep hitting reload" is inevitable, tbh.)

Young fans: if he wants more young fans to get tickets, he could try charging less than £155. Just sayin' like.

[j4.livejournal.com]

They sell out whatever, so why care? Certainly why plough money (& it would be a *lot* of money) into a system that only operates for 2-3 hours every year?

Because a) then it would be guaranteed to work for all the other things seetickets do (and there are other Really Big Instant Sellout gigs, not the same scale, but, y'know) and b) they could then sell their Amazing Totally Un-DDOS-able System to other people and Make $$$$ Fast. Maybe.

Also, c) because otherwise they look monumentally shit once a year, and then for the rest of the year, for the events where they don't have a monopoly on the ticket sales, people are less likely to use them. Again, 'maybe'.

I dunno, I just feel it could be their flagship thing rather than their guaranteed-fuckup thing.

Owen points out that if the tickets were split into different allocations for different resellers eg a quarter to seetickets, a quarter to wegottickets, etc, then a) each of them would only be dealing with a quarter of the problem, and b) there'd be some incentive for them to do it better because of competition.

And, I mean, are we assuming here that Glasto Inc. went out to tender for the ticket-selling contract and seetickets were the best of the bunch? Or are they just Michael Eavis's mates?

[juliet]

But doesn't the same arg apply to Madonna or whatever? (I have 0 idea how the problems compare, mind.) They sell out anyway, so it doesn't matter.

They seem *not* to care about looking like idiots once a year, so I can only assume that it doesn't impact on their business. I agree that it could be flagship rather than fuckup, but from a business POV, that's only worth it if the benefits outweigh the costs. Given that the ticket-selling business seems to be neatly sewn up between about 3 large companies now, the margin of available-business may be quite small.

Sorry, I have my "cynical" hat firmly on. From a more generous POV, I suspect that Glastonbury is actually the biggest deal available by quite a long way in the ticket-selling stakes. So basically they only have a really decent test once a year. They have changed things every year to try and avoid the previous year's cockup; thereby introducing another cockup. I really don't know how difficult or expensive it is to do proper full-on testing: several hundred thousand people all with >1 window open hitting refresh every couple of seconds. You can't do it properly internally because some of the issue is to do with the pipes to the outside world.

Um. It does seem like it *should* be solvable, though.

Seetickets: do they have a deal with the Mean Fiddler?

[barnacle.livejournal.com]

Whenever he appears in the press, Eavis disgusts me, the upsidedownheaded parsimonious fucking farmer that he is. He treats his paying customers like cattle, presumably that being the only mode of interaction that he's happy with. The poor buggers paid through the nose for the privilege to be used last year (or was it the one before?) to test his ace new drainage systems, presumably because it would cost too much to test them properly before the actual event. It still cocking flooded.

This year the organizers were whining that they hadn't had enough interest to ensure a sell-out, prior to their systems going live and promptly falling over. Maybe for one year they might not see the unholy, undignified, mean-spirited scrum that they get every year, and they start complaining. It's pathetic.

Glastonbury seems to be run entirely by greedy, ignorant money-oriented fools who won't spend a pound to gain a hundred. I'd argue that the best thing to do, if you genuinely love the festival's long-term vibe, is to simply not go, until everyone currently involved is bankrupt. Financially, obv.

[uon.livejournal.com]

I don't think boycotting Glastonbury will have any effect on the long-term vibe of the festival.

If everybody who gave a toss stopped going, then it would wind up full of people who didn't give a toss and enjoyed shouting "BOLLOCKS!" at top of voices all night long in the camping areas. They could rename it the Pilton Mean Fiddler Festival and be done with it. Anyone who gave a toss about the vibe, meanwhile, would just go to other festivals; to the extent to which the long-term vibe survived, it wouldn't be in Glastonbury.

(I was impressed that the organisers of this year's Glade festival said they were deliberately reducing the number of tickets, since they thought last year's was a smidge too big.)

I'm quite happy to give it at least one last try this year. The Green Fields are still there, and there's a tiny chance it might not even rain as much as the last two years!

[venta.livejournal.com]

...to test his ace new drainage systems, presumably because it would cost too much to test them properly before the actual event

So, out of interest, how would you simulate four days' rain with 200,000 people walking round the farm for testing purposes ?

[kaet.livejournal.com]

I don't think the technology comes into it much really. These venue ticketing type things are all about power politics and backhanders and dodgy practices and the like. It's a complete shark-pool.

I think I'd probably get around the transaciton problem by getting people to pre-register in a period of a month or so in advance with card details, etc. Then you can distribute count (1000s, say) to a number of distributed servers, which don't communicate except to get another count. The distributed servers just take the id and then collate it for charging, etc, at a later time. The credit transactions which eventually fail release "returns" to the servers. The ids are distbiuted reliably to a particular server, so ids are reliably removed when you get upto the limit. You could run lighttpd and a simple db-less backend.