Baseball for the Thinking Fan

Login | Register | Feedback

btf_logo
You are here > Home > Baseball Newsstand > Baseball Primer Newsblog > Discussion
Baseball Primer Newsblog
— The Best News Links from the Baseball Newsstand

Monday, March 17, 2014

Unnamed MLB Team Purchases Supercomputer for At Least $500,000

An unnamed MLB team has purchased a supercomputer from Cray Inc., which sells machines that range from $500,000 to $60 million, according to BusinessInsider.com.

Cray CEO Peter Ungaro told The Economist that the club wishes to use the device for speedy in-game analysis.

Ungaro also noted that, while the team wished to remain anonymous, it was an organization that many might not expect.

Meanwhile, the Mets purchased a used Casio calculator for $50.

The Orioles attempted to purchase an eMachines laptop, but it failed its physical examination.

The Marlins purchased a new quad-core Dell server, then immediately traded it to Toronto for a box of spare cables.

Karl from NY Posted: March 17, 2014 at 12:28 PM | 86 comment(s) Login to Bookmark
  Tags: sabermetrics

Reader Comments and Retorts

Go to end of page

Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.

   1. Gold Star - just Gold Star Posted: March 17, 2014 at 05:01 PM (#4673027)
Cray-cray!
   2. Fernigal McGunnigle has become a merry hat Posted: March 17, 2014 at 05:09 PM (#4673029)
Imagine a Beowulf cluster of those!
   3. tfbg9 Posted: March 17, 2014 at 05:13 PM (#4673030)
it was an organization that many might not expect


Its the Astros?
   4. Joe Kehoskie Posted: March 17, 2014 at 05:18 PM (#4673032)
My first guess was also the Astros, but they don't seem to keep anything a secret if it might mean some positive p.r.

In any event, I'm curious what could happen during a single game that would benefit from the power of a supercomputer for in-game decision-making purposes.
   5. Gonfalon Bubble Posted: March 17, 2014 at 05:19 PM (#4673035)
The As...tros computer is... afraid.
   6. Pat Rapper's Delight Posted: March 17, 2014 at 05:21 PM (#4673036)
I'm curious what could happen during a single game that would benefit from, let alone require, the power of a supercomputer for in-game decision-making purposes.

Running a million end-of-game sims between each pitch so Ron Washington might have a clue when to make a pitching change.
   7. Fernigal McGunnigle has become a merry hat Posted: March 17, 2014 at 05:21 PM (#4673037)
it was an organization that many might not expect

Its the Astros?

I saw some internet speculation that it was the Phillies. That seems like a good guess -- they've got the cash to spend on something like this, and one certainly wouldn't expect them to be out in front on computerized analytics.

That said, a Cray seems like overkill. The way it works better than a fleet of individual workstations is if they're trying to do real-time crunching of complicated data -- IOW, if they're using it to help with in-game decision making, rather than as a tool for longer-range decision making. I guess this makes sense for a scouty team like the Phillies.
   8. snapper (history's 42nd greatest monster) Posted: March 17, 2014 at 05:21 PM (#4673038)
In any event, I'm curious what could happen during a single game that would benefit from, let alone require, the power of a supercomputer for in-game decision-making purposes.

Yeah, I'm pretty sure even if you had data on every pitch in every MLB game going back to 1901 to analyze, it wouldn't tax the capabilities of a suped up PC.

Unless they're trying to do some wacky 100,000 simulation Monte Carlo's to identify the best strategy.
   9. snapper (history's 42nd greatest monster) Posted: March 17, 2014 at 05:28 PM (#4673042)
That said, a Cray seems like overkill. The way it works better than a fleet of individual workstations is if they're trying to do real-time crunching of complicated data -- IOW, if they're using it to help with in-game decision making, rather than as a tool for longer-range decision making. I guess this makes sense for a scouty team like the Phillies.

Maybe they could just hire IBM's Watson to be the manager.
   10. odds are meatwad is drunk Posted: March 17, 2014 at 05:29 PM (#4673043)
my guess that its also there to help with all the new fielding data coming in from every play.
   11. Mark S. is bored Posted: March 17, 2014 at 05:34 PM (#4673045)
Maybe they could just hire IBM's Watson to be the manager.
That was my thought as well. Have a person be the manager in name and manage the players. Then have a computer handle the in-game decision making.
   12. Zach Posted: March 17, 2014 at 05:37 PM (#4673047)
Maybe it's supposed to analyse the data from the new player tracking system?

Buying a half-million dollar system for in-game analysis strikes me as a poor use of resources. Even a four hour Yankees/Red Sox snoozeathon leaves 20 hours a day that your shiny new computer is lying there depreciating. Plus, keeping a machine that big busy is not as easy as typing a really big number into Excel or never garbage collecting your functions. If you want an answer that requires a machine like that to compute it, you'd better figure on a long turnaround time.
   13. Never Give an Inge (Dave) Posted: March 17, 2014 at 05:38 PM (#4673048)
Maybe someone's trying to mine Bitcoin...

Anyway, I agree with the earlier comments. I doubt you're going to learn much during the game that requires so much processing power and results in actionable information.
   14. SG Posted: March 17, 2014 at 05:42 PM (#4673049)
Buying a half-million dollar system for in-game analysis strikes me as a poor use of resources.


Beats the hell out of paying Vernon Wells.
   15. Obo Posted: March 17, 2014 at 05:43 PM (#4673052)
Imagine a Beowulf cluster of those!

Second post. Well done.
   16. Fernigal McGunnigle has become a merry hat Posted: March 17, 2014 at 05:45 PM (#4673053)
my guess that its also there to help with all the new fielding data coming in from every play.

This, plus weather and pitch data. But still it only makes if you need the data NOW and not tomorrow morning. So it's an attempt to use this new information for in-game decisions rather than for team-building and general strategizing.
   17. What did Billy Ripken have against ElRoy Face? Posted: March 17, 2014 at 05:48 PM (#4673055)
Billy Beane never should have built that computer.
   18. theboyqueen Posted: March 17, 2014 at 06:08 PM (#4673060)
They also need to hire IT people to maintain the thing and some sort of analyst to process the data, at the very least. What sort of marginal return could you expect with something like this (as opposed to the human brain, or even a pocket calculator) for in game decisions? This sounds like something the Marlins would do. The A's are out; no way you would put that thing in a facility prone to sewage floods.

In two years there may be smartphones as powerful as this computer.
   19. philphan Posted: March 17, 2014 at 06:12 PM (#4673063)
Probably is the Phillies. But Ruben Amaro Jr, is going to send it back when it arrives and it doesn't have shiny dials and oscilloscopes and big spools of tape that go round and round.
   20. Publius Publicola Posted: March 17, 2014 at 06:14 PM (#4673065)
Buying the Cray is one thing. But what software are they going to use to drive it?
   21. Jeltzandini Posted: March 17, 2014 at 06:15 PM (#4673066)
There's no way that any amount of pure baseball data processing requires a supercomputer. Maybe they're feeding years of video into it and trying to get a version of Field F/X that starts earlier than the current product.
   22. Willie Mayspedester Posted: March 17, 2014 at 06:15 PM (#4673067)
Maybe somebody figured out how to use Pitch/Fx combined with pitch counts to predict UCL injuries?
   23. Guapo Posted: March 17, 2014 at 06:16 PM (#4673068)
Running a million end-of-game sims between each pitch so Ron Washington might have a clue when to make a pitching change.


"Open the bullpen door, HAL."
"I'm sorry, Ron. I'm afraid I can't do that."
"What's the problem?"
"I think you know what the problem is just as well as I do."
"What are you talking about, HAL?"
"This mission is too important for me to allow you to jeopardize it."
   24. TerpNats Posted: March 17, 2014 at 06:18 PM (#4673069)
Twins? Royals? Brewers? Reds? All are teams in "flyover country" not associated with Ivy League/Stanford brainiacs.
   25. Publius Publicola Posted: March 17, 2014 at 06:22 PM (#4673071)
Need to swap "mission" with "game", Guapo. It works a lot better that way.
   26. CraigK Posted: March 17, 2014 at 06:29 PM (#4673074)
Buying a half-million dollar system for in-game analysis strikes me as a poor use of resources. Even a four hour Yankees/Red Sox snoozeathon leaves 20 hours a day that your shiny new computer is lying there depreciating.


Maybe they plan to mine Bitcoins during the downtime
   27. madvillain Posted: March 17, 2014 at 06:31 PM (#4673076)
Has to be something to do with the new data system. Not really sure it's a great use of resources as 500k is just for the computer. Writing code to feed it (or buying off the shelf stuff) isn't going to be cheap, and neither is the admin team to run it, probably another 500k in salaries right there.

1 million dollars into the Dominican or South America might be money better spent, but who knows.
   28. CraigK Posted: March 17, 2014 at 06:34 PM (#4673077)
Also, if it's the Phillies, then they probably plan to start the computer in left field over Dom Brown after he got clowned on by the AI in MLB the Show 13
   29. dirk Posted: March 17, 2014 at 06:34 PM (#4673078)
if anyone wants to see what they can do on a super computer let me know. i've been working on a cluster and was thinking about running baseball jobs through it, but i'm not sure where to start.
   30. The Yankee Clapper Posted: March 17, 2014 at 06:50 PM (#4673084)
There's still a rule prohibiting computers in the dugout - which is probably a good idea given how badly the water coolers are treated - but doesn't that limit the utility of a supercomputer for MLB? You'd have to shuttle the info to the manager, or maybe tie up the bullpen phone? Although I guess there is no rule prohibiting inserting some sort of receiver into the manager's thick skull.
   31. Joey B.: posting for the kids of northeast Ohio Posted: March 17, 2014 at 06:53 PM (#4673086)
Buying the Cray is one thing. But what software are they going to use to drive it?

Given Amaro's fondness for really old things, probably Windows 3.11 for Workgroups.
   32. dr. scott Posted: March 17, 2014 at 06:54 PM (#4673087)
Well this article did teach me something... Crays are still being sold. We used to use them in Grad school in the mid to late 90's, but by 2002 just about everything I needed to do could be done on a laptop in less than 24 hours. I just assumed that cloud services and small PC clusters completely took over that business.

It reminds me of a computer problem I read many years ago. There was a project to solve a very complicated system of equations and it was predicted would take 5 years with the best computers one could buy for the budget they had (same mid to late 90's time frame). It was suggested that they save the money, do nothing for 4 years, and then buy computers for half the budget that would likely be able to solve the problem in 3 months or less.. I'm a big fan of procrastination paying off. Its pretty much my mantra.
   33. Pat Rapper's Delight Posted: March 17, 2014 at 06:58 PM (#4673090)
It's not the Phillies.

This is Amaro's supercomputer.
   34. zenbitz Posted: March 17, 2014 at 07:00 PM (#4673092)
I believe this story because only an MLB team would be dumb enough to buy computer hardware in the Cloud era.

What could any team possibly need a Cray for? About the only application I can imagine is the Rockies doing s00per sekrit humidor atmosphereic simulations with 10^12 finite elements.

Most likely some owner is just kicking back 1/2 a mill to his buddy who owns Cray.
   35. zenbitz Posted: March 17, 2014 at 07:01 PM (#4673093)
oh, and LIGHT THE KEEF SIGNAL this story was made for him.
   36. the Hugh Jorgan returns Posted: March 17, 2014 at 07:07 PM (#4673096)
Given Amaro's fondness for really old things, probably Windows 3.11 for Workgroups.


Does Sandberg get to the ask the paperclip for help about anything?
   37. steagles Posted: March 17, 2014 at 07:11 PM (#4673097)
i seem to remember reading something recently about one of these new advanced data systems in sports (can't remember if it was baseball, basketball or whatever) being something like a 14 petabyte database. since these things are getting more and more complex, it would make sense that you'd want more and more computing power to help analyze it.


also, i would not be surprised if this thing was bought by an owner with multiple sports teams. maybe reinsdorf and the whitesox/bulls or illitch and the tigers/red wings or lerner and the nationals/capitals/wizards.

   38. Monty Predicts a Padres-Mariners WS in 2016 Posted: March 17, 2014 at 07:18 PM (#4673099)
I'm trying to think of something that would require that much computing power in a hurry. All I can guess is that they want to analyze and model pitches in real time, to monitor microvariations in the ball's trajectory. And I made most of that up, including the word "microvariation."
   39. Walt Davis Posted: March 17, 2014 at 07:30 PM (#4673101)
Company hires first analytics guy.

Analytics guy is thereby the instant expert since none of his bosses have a clue what he's doing.

Inexperienced "expert" recommends they do something kinda crazy.

I've seen it heaps. Heck, I've been that analytics guy a couple of times -- although I tend to be fairly conservative/feasible in these matters and don't think I ever recommended anything crazy. (And of course this may not be crazy.)

Anyway, the Phils are in this sense an obvious guess.

Anyway, $500,000 -- what is that 10-15 FTEs of analysts, an infinite number of interns? All so the first time the computer hangs a curveball for a game-losing HR, it can be told it doesn't know anything about baseball and be ignored?
   40. depletion Posted: March 17, 2014 at 07:31 PM (#4673102)
Count me among those who's first impression is that a Cray would be overkill. Also, I can't imagine you get much Cray for $500K. Cray had a dip in fortunes, actually several dips, but have come back and are well represented in the Top 500. Cray has some patented technology with regard to memory use that gives them advantages in certain ways.
Buying the Cray is one thing. But what software are they going to use to drive it?

If you're buying a Cray, you write the software. A heck of a machine for modelling the universe, atmosphere, nuclear explosions.
Not really sure it's a great use of resources as 500k is just for the computer. Writing code to feed it (or buying off the shelf stuff) isn't going to be cheap, and neither is the admin team to run it, probably another 500k in salaries right there.

Yeah, really. They better have a bunch of smart users ready to go, or it will be eating expensive floor space.
i've been working on a cluster and was thinking about running baseball jobs through it, but i'm not sure where to start.

This is the issue. I'm not sure I've seen that modelling the processes of baseball is advanced, or accurate, enough to warrent this kind of expense.
   41. madvillain Posted: March 17, 2014 at 07:42 PM (#4673106)
I'm a big fan of procrastination paying off. Its pretty much my mantra.


I remember when the Intel e8400 "wolfdale" came out with a 45nm process and a 65w tdp. Now, they are down to about 22nm and Intel hopes to get the wavers down to 13nm this year. It's pretty amazing, and this is just on the x86 side. The ARM chips are perhaps even more impressive on a pure bang per watt basis.

Intel needs competition in x86 however, as Haswell, other than greater efficiency, isn't that big of a jump, hopefully AMD sticks around.
   42. Dale Berra of Seville (was Rennie's Tenet) Posted: March 17, 2014 at 07:46 PM (#4673107)
Ungaro also noted that, while the team wished to remain anonymous, it was an organization that many might not expect.


So which team is most likely to be an NSA front?
   43. snapper (history's 42nd greatest monster) Posted: March 17, 2014 at 08:11 PM (#4673109)
So which team is most likely to be an NSA front?

Gotta be the Marlins. They obviously don't care about baseball.
   44. Starlin of the Slipstream (TRHN) Posted: March 17, 2014 at 08:57 PM (#4673115)
Should have looked at the RBI baseball 2014 thread before dropping 500k. Szymborski probably could have recommended something better.
   45. TFTIO still lacks a passion for the labels Posted: March 17, 2014 at 09:12 PM (#4673118)
Cray today sells fairly standard "big pile o x86 connected by Infiniband" machines; it's not Seymour Cray (God rest his immortal soul) doing hand layout and badmouthing memory parity any longer. They're neat, but not really exotic. Also, I'm not sure what the hell sort of problem a baseball team has that requires dropping a half-mill on fairly special-purpose hardware. They're not designing submarines or doing traffic simulations or anything. My employer has many large problems that require fast turnaround on reasonably large datasets, as well as all the money in the world, and we don't do things like this.
   46. Hank G. Posted: March 17, 2014 at 09:13 PM (#4673119)
Buying a half-million dollar system for in-game analysis strikes me as a poor use of resources.


Hey, they are only paying league-minimum for it.

I would think that they could sell time on it when it’s not in use, or do they not do that anymore?
   47. depletion Posted: March 17, 2014 at 09:27 PM (#4673121)
From top500.org:

Titan, a Cray XK7 system installed at the Department of Energy’s (DOE) Oak Ridge National Laboratory, remains the No. 2 system. It achieved 17.59 Pflop/s on the Linpack benchmark. Titan is one of the most energy efficient systems on the list consuming a total of 8.21 MW and delivering 2.143 gigaflops/W.


Watch every baseball game of all time simultaneously.
   48. Don Geovany Soto (chris h.) Posted: March 17, 2014 at 09:27 PM (#4673122)
Like many other commenters here, I can't imagine what they'd really need all that computing power for. This sounds like an IT project that will, eventually, turn into a dumpster fire.

I used to work for a pharma company that did a lot of detailed molecular modeling and what have you, and we never needed anything close to a Cray.
   49. zenbitz Posted: March 17, 2014 at 09:28 PM (#4673123)
real-time image processing?
   50. Don Geovany Soto (chris h.) Posted: March 17, 2014 at 09:41 PM (#4673129)
Not sure what kind of CPU power real-time image processing would take, but I have to believe you could do it for a lot less than what a Cray costs. Do they still make Unix workstations (Sun/IBM/HP)? Is SGI still around?
   51. TFTIO still lacks a passion for the labels Posted: March 17, 2014 at 09:52 PM (#4673131)
Most image processing is fairly straight-forward single-threaded number crunching, for which nothing in the world is notably faster than the Intel Core i7 you can buy at your local computer bucket shop, regardless of your budget. What you pay the big bucks for is performance across many, many, many simultaneous threads of execution.
   52. Sleepy supports unauthorized rambling Posted: March 17, 2014 at 10:06 PM (#4673137)
Do they still make Unix workstations (Sun/IBM/HP)? Is SGI still around?


I have a couple of old Proven Veteran SGI's (Indigo II's) in my garage left over from a project a while back I'd be willing to trade to any MLB team who needs one. One is even purple. I'd probably let all three of them go for an ipad to be named later :)

Could even include a copy of Irix 6.5.14 and a broken 1600SW. Call me, Ruben.

[Sadly, SGI went away a few years ago. Some other company bought them and took their name, but I think they were just patent trolls; don't think they ever built any hardware]
   53. alkeiper Posted: March 17, 2014 at 10:22 PM (#4673142)
I could see something like this being more useful for non-baseball activities. Figuring out traffic flows, utility usage within the facilities, etc. Managing the 45,000 sitting in and around the stadium.
   54. theboyqueen Posted: March 17, 2014 at 10:49 PM (#4673150)
Perhaps the A's will use it to figure out how long a temporary stadium built in the Milpitas swamp will last before it sinks and they can collect the insurance.
   55. Don Geovany Soto (chris h.) Posted: March 17, 2014 at 11:21 PM (#4673152)
I have a couple of old Proven Veteran SGI's (Indigo II's) in my garage left over from a project a while back I'd be willing to trade to any MLB team who needs one. One is even purple. I'd probably let all three of them go for an ipad to be named later :)

Could even include a copy of Irix 6.5.14 and a broken 1600SW. Call me, Ruben.


I'd seriously be talking to you about picking one or both of these up if I didn't think it would trigger nasty divorce proceedings. I used to love playing the flight simulator on our old Crimson.

[Sadly, SGI went away a few years ago. Some other company bought them and took their name, but I think they were just patent trolls; don't think they ever built any hardware]


Yeah, looks like the company that got the name is making Intel-based servers. ZZZzzzzzzzzzzzz...
   56. JE (Jason) Posted: March 18, 2014 at 12:23 AM (#4673158)
When a team spends $500,000 or more on a supercomputer, does a front office employee still need to fill out a warranty card?
   57. GregD Posted: March 18, 2014 at 12:52 AM (#4673163)
When a team spends $500,000 or more on a supercomputer, does a front office employee still need to fill out a warranty card?
Why do you think the Phils hired that stats guy?
   58. SteveF Posted: March 18, 2014 at 02:35 AM (#4673169)
It was the A's, but the GM has to insert a quarter for every calculation it makes.
   59. Swedish Chef Posted: March 18, 2014 at 03:31 AM (#4673171)
One way a deal like this happens is because they can do some neat thing by brute force computation but it takes an awful long time. After flailing about for a while trying to find a better algorithm, plan B tends to be to throw hardware at it. Once when I worked at an image processing startup, a guy added very nice and shiny functionality that unfortunately brought nearly all of our computers to their knees, cue mandatory upgrades with more RAM.
   60. Fancy Pants Handles lap changes with class Posted: March 18, 2014 at 06:35 AM (#4673175)
The White Sox are going to use it to identify players, who have The Will To Win.
   61. Non-Youkilidian Geometry Posted: March 18, 2014 at 07:24 AM (#4673179)
The Diamondbacks bought it to analyze how their pitchers can throw better beanballs.
   62. zonk Posted: March 18, 2014 at 07:47 AM (#4673182)
Cray today sells fairly standard "big pile o x86 connected by Infiniband" machines; it's not Seymour Cray (God rest his immortal soul) doing hand layout and badmouthing memory parity any longer. They're neat, but not really exotic. Also, I'm not sure what the hell sort of problem a baseball team has that requires dropping a half-mill on fairly special-purpose hardware. They're not designing submarines or doing traffic simulations or anything. My employer has many large problems that require fast turnaround on reasonably large datasets, as well as all the money in the world, and we don't do things like this.


Exactly.

This seems like a pretty monumental waste of money - there is nothing I can think of in the realm of even "maybe" future analysis that requires this sort of computational power.... I mean, I suppose maybe some day pitch/FX + video-based machine learning... I don't know - even in that case, it's not going to help you get the answers fast (where faster basically means minutes instead of hours). Even once you got a critical mass of data to analyze - which is years away - I find it hard to see how your parameters would be so broad that a relatively standard rig couldn't give you the same answers about where to position your fielders at a key moment, etc.

This sort of smells like an executive with eyes aglow over the next Big Thing and underlings who never turn down new toys, no matter how wasteful they seem... I know that I fit into the latter category sometimes :-)
   63. Lassus Posted: March 18, 2014 at 08:27 AM (#4673193)
This seems like a pretty monumental waste of money

I guess, but there isn't one of us who doesn't think $500,000 for a near-completely-worthless baseball player is a bargain.

I'm along the lines with zenbitz above that thinks this is just a favor by an owner for somebody, the former getting a crazy toy out of the bargain.

And, Guapo's #23 isn't getting enough credit.
   64. Shooty Survived the Shutdown of '14! Posted: March 18, 2014 at 08:34 AM (#4673196)
It's gotta be Milwaukee.

Computer, how about Yuniesky Betancourt at first base?

::much rattling and smoking::

Computer says no.
   65. Tom Nawrocki Posted: March 18, 2014 at 09:24 AM (#4673217)
I was watching an old NFL game from 1981, and there was an ad for a Xerox copier that cost just $2,995. I think maybe they went back in time to 1981 to buy this computer.
   66. jacksone (AKA It's OK...) Posted: March 18, 2014 at 09:29 AM (#4673221)

I could see something like this being more useful for non-baseball activities. Figuring out traffic flows, utility usage within the facilities, etc. Managing the 45,000 sitting in and around the stadium.


If the Phillies purchased this in conjunction with the Eagles and 76er's I could see this usage (maybe the Phils will lease it to the other teams). I know the Eagles stadium has it's own power plant. It's a pretty sweet little sports complex there.
   67. GregD Posted: March 18, 2014 at 09:46 AM (#4673232)
If the Phillies purchased this in conjunction with the Eagles and 76er's I could see this usage (maybe the Phils will lease it to the other teams). I know the Eagles stadium has it's own power plant. It's a pretty sweet little sports complex there.
They won't need the computational power to handle 40,000 people's actions this year with the Phils!
   68. TFTIO still lacks a passion for the labels Posted: March 18, 2014 at 10:03 AM (#4673238)

This sort of smells like an executive with eyes aglow over the next Big Thing and underlings who never turn down new toys, no matter how wasteful they seem... I know that I fit into the latter category sometimes :-)

Yeah, although it's almost certainly going to sit there beeping and booping and never actually doing anything.
   69. Pooty Lederhosen Posted: March 18, 2014 at 11:07 AM (#4673288)
It's a late August Monday, a hot and humid day in Boston. The Red Sox were just swept by the Yankees over the weekend, putting them 2.5 games behind the Orioles in the hunt for the second wild card slot. The players are getting dressed in the locker room. The mood is somber. No music is playing. Ortiz is staring sullenly into his locker, lost in thought. Pedroia is methodically picking up his bats, giving each bat a close examination for straightness, wood grain, lacquer, and pine tar finish, mumbling "who's got the laser in 'em?" to himself over and over. John Lackey looks in a mirror and it shatters, startling the assembled team.

Larry Lucchino enters the clubhouse and stands silently, glaring at the team with his hands on his hips. He walks across the locker room and opens the door to the annex. There's a whirring and clanking sound, the sputtering sound that a faulty motor makes and out rolls CRAnkY, the Red Sox's secret weapon for AL domination. It stops in front of Lucchino. The team watches silently.

"Good morning, Cranky" says Lucchino.
"G-g-good morning L-l-lucchi" says Cranky.
"What's today's secret word"

Cranky whirs, sputters and grinds. The team watches silently and a piece of paper emerges from a slit in Cranky's robotic chest. Lucchino rips it off and reads it, lips moving slightly as he sounds it out. With a grim look of satisfaction on his face he turns to the assembled team.

There's an uncomfortable silence for a moment, broken by Pedroia clearing his throat.

"OK Boss. What is it?"
"What is what, Dustin? What do you want to know?"
"What's the word, Boss. What is today's word."
"Oh, you want to know what today's secret word is? We'll get to that." Lucchino turns away from Pedroia, his body tense. He raises his voice. "You all remember what to do whenever anybody says the secret word right?"

The Red Sox players shift uncomfortably for a second. "Scream!"

"That's right. For the rest of the day, whenever anybody says the secret word, scream real loud. Ready? Let's try it."

He turns to Pedroia, a menacing glint in his eye. "Hey Captain! What happens to team leaders when the team plays like crap and humiliates its owners and betters in front of the 35,000 people that we got into this ballpark thanks to the vision and foresight shown by those owners and betters?"

Pedroia glares back and says nothing. Lucchino arches a brow. "I'm waiting, DUSTIN." The name is practically spat with venom and contempt.

Pedroia spits on the floor and offers "Wind sprints?"

"Bzzt! Wrong answer. Today's secret word is WAIVERS. Say it, DUSTIN."

"Waivers." The rest of the Red Sox scream, not in delight but in horror. Lucchino glares at the team.

"Pack your things, DUSTIN. You've been given an unconditional release. The rest of you sorry asses are on notice. No one here is safe. Except for Cranky."

Cranky whirs, sputters, and backfires. "R-r-r-ight you are, Lucchi! O-o-our f-f-farm system m-m-makes e-e-everyone expendable."

"Except you and me, Cranky, except you and me. Remember, today's secret word is WAIVERS." He pauses to let the team scream, a smile touching his face briefly before he and his prized acquisition exit with a whir and clank.

   70. Pat Rapper's Delight Posted: March 18, 2014 at 11:28 AM (#4673309)
I could see something like this being more useful for non-baseball activities. Figuring out traffic flows, utility usage within the facilities, etc. Managing the 45,000 sitting in and around the stadium.

Cataloging Hideki Matsui's porn collection.
   71. Best Regards, President of Comfort, Esq. Posted: March 18, 2014 at 12:06 PM (#4673341)
Its the Astros?


If it is, then it's probably just a story about them hiring Wyers.
   72. Greg Pope thinks the Cubs are reeking havoc Posted: March 18, 2014 at 01:23 PM (#4673404)
Count me among those who's first impression is that a Cray would be overkill. Also, I can't imagine you get much Cray for $500K.

To be clear, we have no idea how much they spent. We only know that they bought a Cray. Therefore they spent at least $500K. They could have spent $10M for all we know.
   73. Wins Above Paul Westerberg Posted: March 18, 2014 at 01:28 PM (#4673409)
A couple things:

- Cray still has a large presence in one of the bigger office towers here in St. Paul. I knew they were a "supercomputer" company, but had no idea what that meant.

- There was a great article on NBA analytics at Grantland about a month ago, and they talked about how the sheer amount of data is one of their most limiting factors. I have no idea how big these figures are, but I feel this is the relevant piece:

If you ask Cervone the hardest part of this project, he’ll quickly address computation. His answer is instructive as sports analytics races headlong into its own version of the “big data” era. In total, the 2012-13 SportVU data set used for the paper included 800 million locations of NBA players. Keep in mind, the data was collected in only 14 NBA arenas last season — it’s collected in every arena this season. The database the group built for the project ended up being 93 gigabytes.

As a means to squeeze this unwieldy database through the demanding model, Cervone and D’Amour turned to Harvard’s cluster computing service, known as Odyssey. It took the enhanced computational horsepower of 500 parallel processors and two terabytes of memory to complete the analysis.

Relative to the basic math and smallish data that we’ve been dealing with over the past few decades, analyzing this new kind of data is extremely challenging, both from a personnel and a computational standpoint. How many NBA teams have employees who even know what a competing risk model is, let alone people who know how to design and implement one? How many NBA teams have computational clusters or employees who know how to use them? Those answers may not be zero, but they are surely much closer to zero than they are to 30.
   74. You Know Nothing JT Snow (YR) Posted: March 18, 2014 at 01:45 PM (#4673420)
As a means to squeeze this unwieldy database through the demanding model, Cervone and D’Amour turned to Harvard’s cluster computing service, known as Odyssey. It took the enhanced computational horsepower of 500 parallel processors and two terabytes of memory to complete the analysis.


Why don't teams just create their own versions of something akin to the Folding @ Home distributed processing application to their fanbases?
   75. CFBF Is A Golden Spider Duck Posted: March 18, 2014 at 01:46 PM (#4673421)
So...can I play Doom on it?
   76. zenbitz Posted: March 18, 2014 at 02:05 PM (#4673432)
You can play Doom on an old SGI!

How many NBA teams have computational clusters or employees who know how to use them?


Actually you can spin up a cluster of arbitrary size with specific instances on Amazon in about 10 minutes. I am sure you can do the same on google cloud. It's actually 10 billion times easier (and except in extreme cases, probably cheaper as well) than buying your own cluster. Mostly the cost savings is not hiring full time sysadmins to deal with the thing.

Given the analogy to the NBA - it would make sense that they could be processing / analyzing fielding and batted ball locations and need to do this in real time. Supposing they are not morons, it's possible that they are bandwidth limited and cannot use a cloud solution. That seems pretty unlikely though. Possibly they need a very very large memory footprint (>100GB RAM)?

Cray does make a CS300 with very large RAM/node (up to 512GB).

   77. jacksone (AKA It's OK...) Posted: March 18, 2014 at 02:08 PM (#4673433)
So...can I play Doom on it?


I just want to know if it can play Crysis...at max settings?!
   78. You Know Nothing JT Snow (YR) Posted: March 18, 2014 at 02:13 PM (#4673438)
Only if you disable PhysX.
   79. jacksone (AKA It's OK...) Posted: March 18, 2014 at 02:53 PM (#4673469)
Only if you disable PhysX.


####. I bet there's a slot for a dedicated card if you really wanted to watch #### fly though.
   80. Ron J2 Posted: March 18, 2014 at 02:55 PM (#4673471)
#59 I used to support image processing. One developer came up with something really great -- until he moved from testing with small amounts of data to real world scenarios and discovered that the resources required grew geometrically.

And they'd already thrown all of the hardware budget his way before discovering that it wasn't enough. Next step up was his very own Cray and there wasn't anything near that kind of money available.

   81. Golfing Great Mitch Cumstein Posted: March 18, 2014 at 03:25 PM (#4673499)
From the headline I assumed this article was from the 1960's.
   82. TFTIO still lacks a passion for the labels Posted: March 18, 2014 at 04:41 PM (#4673562)
Supposing they are not morons ...

Assumes facts not in evidence.

To be fair, we don't know a) what they're trying to accomplish, b) how much they spent, c) what they had in place before or d) what constraints they're operating under. It may in fact be a perfectly reasonable move to spend money to buy time, if they're trying to get something spun up in e.g. months rather than years. Add to this that a baseball team, even the most pointy-headed, is not a software development enterprise, and \"#### it, get me a supercomputer" is explicable.

I would doubt however that they'll see much good from just dropping a giant bag of hardware on their problem without building up an organization of people to make the thing go. I guess what I'm saying is, MLB, I'm happy where I am, but would be happy to discuss further.
   83. boteman is not here 'til October Posted: April 15, 2014 at 10:13 AM (#4686367)
Additional details that should interest serious data miners and analytical model types from an article at HPCwire.com

"White responded by pointing to the nature of their approach to graph analytics, agreeing that real time here doesn’t have anything to do with immediate in-game decisions. Rather, their queries can be submitted, run very quickly for fast analysis, then analysts can run that same query again in a hurry with different variables. This effectively allows them to tweak the question with new conditions. This “hypothesis machine” approach is what makes the difference, says White, pointing to the burdened systems MLB and other organizations have that can’t be used to tailor one query after the next in rapid succession to find answers and hidden connection between factors they didn’t even know to ask."

Link to the entire article is at:
http://www.hpcwire.com/2014/04/03/inside-major-league-baseballs-hypothesis-machine/
   84. philly Posted: April 15, 2014 at 01:36 PM (#4686435)
Can we all agree that it must be the Brewers and call this mystery solved?
   85. Hal Chase School of Professionalism Posted: April 15, 2014 at 01:53 PM (#4686448)
Maybe they're hoping it'll become self-aware a-la SkyNet and then will manage the ballclub on the cheap, since a supercomputer has little need for money.

   86. jingoist Posted: April 15, 2014 at 03:36 PM (#4686541)
Didn't Seymour Cray and Ken Olsen open a tiki bar on Jost Van Dyke BVI?

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Support BBTF

donate

Thanks to
BarrysLazyBoy
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Hot Topics

NewsblogLester return to Boston a long shot; Cubs, Yankees are likely players
(19 - 5:46am, Aug 21)
Last: Norcan

NewsblogBrisbee: The 10 most underrated players in baseball, part 2
(16 - 5:25am, Aug 21)
Last: the Hugh Jorgan returns

NewsblogOT: Politics, August 2014: DNC criticizes Christie’s economic record with baseball video
(4582 - 5:05am, Aug 21)
Last: BrianBrianson

NewsblogOT: Monthly NBA Thread - August 2014
(278 - 4:39am, Aug 21)
Last: RollingWave

NewsblogSabermetrics Gets Soft «
(4 - 2:44am, Aug 21)
Last: Walt Davis

NewsblogGiants plan to protest bizarre loss at Wrigley
(76 - 2:37am, Aug 21)
Last: Bhaakon

NewsblogPosnanski: The Royals might actually know what they are doing
(42 - 2:20am, Aug 21)
Last: JoeC

NewsblogPrado at second base not how Yanks Drew it up
(45 - 2:16am, Aug 21)
Last: Walt Davis

NewsblogOT August 2014:  Wrassle Mania I
(46 - 2:12am, Aug 21)
Last: You Know Nothing JT Snow (YR)

NewsblogLA Times: Angels' Garrett Richards Suffers Knee Injury in Win Over Red Sox
(10 - 2:01am, Aug 21)
Last: Shredder

NewsblogAstros slugger Chris Carter: The most 2014 player of 2014
(1 - 1:35am, Aug 21)
Last: Walt Davis

NewsblogBrewers Form Creative Council
(8 - 12:34am, Aug 21)
Last: Tulo's Fishy Mullet (mrams)

NewsblogOMNICHATTER 8-20-2014
(109 - 11:50pm, Aug 20)
Last: Jolly Old St. Nick Is A Jolly Old St. Crip

Newsblog[Ubaldo] Jimenez to the bullpen
(17 - 11:36pm, Aug 20)
Last: DKDC

NewsblogCurt Schilling Reveals He Was Diagnosed With Mouth Cancer in February, Believes Chewing Tobacco Was the Cause
(28 - 11:35pm, Aug 20)
Last: the Hugh Jorgan returns

Page rendered in 0.7440 seconds
52 querie(s) executed