Stats programs and you

[Avram]

At the urging of several people who keep asking me why few people use QBSQL, the stats system that I announced about a year and a half ago, I'll pose the question:
Why do you prefer to use SQBS?

It's been encouraging to see more teams use Taft to simplify their hosting, and recently two more schools have started using QBSQL for their tournaments, but the vast majority of events are run using SQBS.

I feel like I'm trolling or advertising by writing this, but I honestly don't know why tournament directors have been so loyal to SQBS.

If there is the user community to support it, I have realistic ideas for new directions to take in quizbowl stats, but I'd first like to make sure that I'm not overlooking some key functionality that is keeping tournament directors away from trying it out. Taft, QBSQL, and other projects would all benefit from increased innovation and interest in statskeeping.

So if you are someone who sometimes runs stats, take a look at the test server, create a tournament to play around with, upload some old SQBS data, and let me know what you think.

[dtaylor4]

I stick with it because I'm very comfortable with it.

Also, it can be used without an internet connection. This is less of an issue than it used to be, but wireless connections are not 100% reliable.

[Avram]

Also, it can be used without an internet connection. This is less of an issue than it used to be, but wireless connections are not 100% reliable.

I wish I could say that I had a good solution to this, but I don't. This is one of the reasons I added SQBS import and export, so that I can switch systems if necessary, but that's still not what you want to be doing mid-tournament. NAQT's server-based system is made reliable by having the server local-- right there in the stats room (I think).

[Mechanical Beasts]

Also, it can be used without an internet connection. This is less of an issue than it used to be, but wireless connections are not 100% reliable.

I wish I could say that I had a good solution to this, but I don't. This is one of the reasons I added SQBS import and export, so that I can switch systems if necessary, but that's still not what you want to be doing mid-tournament. NAQT's server-based system is made reliable by having the server local-- right there in the stats room (I think).

Avram, couldn't someone just run Apache off their computer and install QBSQL on that?

[The Toad to Wigan Pier]

You could make a virtual appliance of it and run it that way which I imagine is easier for non-technical people than installing apache+friends.
EDIT: Spelling

[Avram]

Avram, couldn't someone just run Apache off their computer and install QBSQL on that?

Yes, but not many hosts want to get into that kind of set-up.

[Mechanical Beasts]

Avram, couldn't someone just run Apache off their computer and install QBSQL on that?

Yes, but not many hosts want to get into that kind of set-up.

Then how do many hosts learn PHP?

Oh.

[Sen. Estes Kefauver (D-TN)]

Even if I could figure out how to use your program, I wouldn't because I think it looks horrible compared to SQBS's output.

[Mechanical Beasts]

Even if I could figure out how to use your program, I wouldn't because I think it looks horrible compared to SQBS's output.

I mean, if you were to set it up on your server/ not on gimranov or whatever, it wouldn't have to have Avram's style sheet.

[Matt Weiner]

Well, TAFT doesn't really "work" per se for large tournaments and I don't trust other people to host my stats in perpetuity. If QBSQL solved these issues I would certainly try it out, but I've used SQBS for a long time and I know how to make it do what I want, so I don't tend to fix what isn't broken.

[dtaylor4]

Despite praising SQBS, I am actually going to test Gimranov in a couple weeks at NAQT State. If I am able to work it to my needs, it will be given a true test at the Barrington MS tournament.

The main issue I have run into with SQBS is that only one machine can input data. If there is a logjam with regards to inputting data (see my failed attempt to do live stats for the Springfield tournament), it is almost impossible to find a quick fix.

[Avram]

Even if I could figure out how to use your program, I wouldn't because I think it looks horrible compared to SQBS's output.

I encourage you to go to http://quizbowl.gimranov.com/qbsql/index.php?t=test_sct, log in with the username "test" and password "test", and click on "Add a game" at the bottom of the page. I believe the system is fairly intuitive for users or people entering stats at least-- please let me know if you get confused by something.

Do you prefer the table styles of SQBS output, or the fact that SQBS puts all the team summaries and player summaries on a single page? The first could be easily be changed via user stylesheets-- Berry College has made some simple changes of that sort. The second is more a design choice-- since QBSQL doesn't try to fit all of its output into 4 webpages, it separates out players, teams, and rounds onto separate pages. If there's a good reason to make four pages that include everything a la SQBS, that would be very easy to do.

Well, TAFT doesn't really "work" per se for large tournaments and I don't trust other people to host my stats in perpetuity.

I just noticed the brackets issues that Taft has. QBSQL's treatment of brackets is less than perfect, but it is the opposite of Taft's; QBSQL doesn't yet know how to show brackets that aren't ranked, so if you bracket teams, the teams in the top bracket will necessarily be listed above those in the second bracket, etc. Take a look at the D2 stats from this year's SCT to see how I handle it.

As for hosting in perpetuity, I don't promise to host your stats in perpetuity, to be honest. I do promise that my code will be available in perpetuity, and I provide a means of downloading the stats in a useful format (SQBS format) whenever you like. Export as several HTML files is planned (see above), but I haven't done it yet, mainly because I haven't seen much need.

[Avram]

You could make a virtual appliance of it and run it that way which I imagine is easier for non-technical people than installing apache+friends.

My thinking is that some programs will want to run their own copies, and they can find their one tech-oriented person who can upload a folder of scripts and set up the database; from then on out, it should be a turnkey operation, with no need for technical expertise.

I don't know much about creating virtual appliances, and I don't see the need, but I'd be glad to work with someone else who thinks it'd be useful to make such a distribution of QBSQL.

I can also host tournaments for teams on my server at http://quizbowl.gimranov.com, which I'm glad to do for free. If teams want to do this, they need only email me and ask for me to create a tournament. As I said in my response to Matt, I certainly won't promise that I'll still pay for that server 15 years from now, but there is robust export facility to SQBS format (and eventually to some other more transparent formats), so your data will not be lost.

[jonpin]

I'm going to agree with Charlie: the output looks poor compared with SQBS. Each game has its own page, which could be cool in principle, but I'm not convinced that's necessary unless a play-by-play is included. Otherwise, QBSQL takes a separate page to show what SQBS displays in four lines. Further, while this is something that can be changed fairly simply I imagine, the grand displays are not well chosen. For an individual team's page, the only data shown for each game is the score, each team's bonus conversion, and the completely superfluous margin-of-victory. Using this page to look up data is awful, and when I was putting together my D-value projections, I just abandoned Region 15 until NAQT posted the data on their site in pseudo-SQBS display form, because I didn't want to have to click a separate page to find out P/T/N for each game. Even a team's bonus conversions for each game aren't all aligned, because they're in a different place when the team lost than when they won.

Furthermore I'm going to agree with Matt: SQBS works. When Bergen was getting set to host a tournament last spring, I told the appointed statkeeper to download SQBS and play with it for five minutes. That's all it takes to understand how to do just about everything it does. Yes, it bottlenecks, but that doesn't matter if your tournament has less than 40 teams.

[the return of AHAN]

Is there a specific program my statkeeper should be using if he's going to try to track 48 teams playing in 2 separate but parallel tournaments?

[Joe Romersa]

Is there a specific program my statkeeper should be using if he's going to try to track 48 teams playing in 2 separate but parallel tournaments?

Have two stats people?

[Avram]

I'm going to agree with Charlie: the output looks poor compared with SQBS. Each game has its own page, which could be cool in principle, but I'm not convinced that's necessary unless a play-by-play is included.

The game detail was not intended to replace the grand summaries-- I'll improve the summaries as you recommend. The detail provides, I think, a good look at how a game played out on a player-by-player basis. I do plan on adding play-by-play, which will of course be added to this page.

For an individual team's page, the only data shown for each game is the score, each team's bonus conversion, and the completely superfluous margin-of-victory. Using this page to look up data is awful, and when I was putting together my D-value projections, I just abandoned Region 15 until NAQT posted the data on their site in pseudo-SQBS display form, because I didn't want to have to click a separate page to find out P/T/N for each game. Even a team's bonus conversions for each game aren't all aligned, because they're in a different place when the team lost than when they won.

These are, as you note, are easy to change. I'll see about adding these in the near future.

Now that play-by-play has been broached as a topic, I'd like to raise a hypothetical-- should tossup-level data be provided before a set has been released for discussion? Knowledge of, say, who powered tossup 10 in packet 4 might be useful in some cases.

[Avram]

Is there a specific program my statkeeper should be using if he's going to try to track 48 teams playing in 2 separate but parallel tournaments?

QBSQL handles this well-- I ran D1 and D2 stats simultaneously for the Region 15/West SCT this year and last without a problem. I created two tournaments and added the games each to the appropriate tournament. If you can't handle this with two divisons in one SQBS file, you may want to simply find a second stats computer.

[the return of AHAN]

... should tossup-level data be provided before a set has been released for discussion? Knowledge of, say, who powered tossup 10 in packet 4 might be useful in some cases.

OOOH... I don't think so. I'm not sure anyone in a match should know the conversion rate for any given toss-up before it's heard. I'd think it'd make you play a TU one way if you knew it has had 100% conversion, while you'd handle it much differently if you knew it's conversion rate was, say, below 50%, and especially if you could a see a stat showing the first buzz typically negged it.

[Whiter Hydra]

Well, isn't BEeS supposed to come out soon?

[Auks Ran Ova]

Well, isn't BEeS supposed to come out soon?

They haven't rounded up all the unicorns necessary to start work yet.