I agree with Matt that it's non-linear at its endpoints. We debated this a couple months ago on Lookout Landing, but reaching the conclusion required a better understanding of wOBA than most of us had at the time... I hit up a fellow SABR guy (and friend of mine, and current MIT student) for his opinion and he convinced me.
My understanding, which is derived from my MIT friend's understanding, is that as long as you have at most ONE black hole in your lineup team wOBA will accurately correlate with runs scored. However, if you have TWO black holes close together in your lineup (read: Brendan Ryan and Robert Andino), the correlation starts to break down. This is because of the way that wOBA works. The linear weights for events used in wOBA are generated by taking the difference in run expectancy between the initial state and the final state; for instance, if the run expectancy with no one on and no one out is 0.4, and the run expectancy of a man on first with no one out is 0.7, then the value of a leadoff single is 0.3 runs. wOBA then averages the values of all singles, or doubles, or triples, or whatever, to get linear weights for each event. Therefore, they pretty obviously accurately represent the value of setting up for the next guy (or, in Brendan Ryan's case, failing to set up for him). However, when you have two terrible hitters in a row, not only are you failing to set up for a guy, but he's taking your bad setup and failing even further. wOBA doesn't account for that double-failure. It thinks that the good hitter after those terrible two will have a normal chance of getting RBIs with hits, which obviously he doesn't.
Since OPS is just a less good version of wOBA (and OPS+ a less good version of wRC+), it has the same flaws. The other reason, of course, that the Mariners' OPS+ never correlated with their runs scored is because OPS+ is park adjusted and runs scored is not.
Seriously, though: one of my few SSI pet peeves is that the conversation is always in terms of OPS and OPS+, which to my mind are pretty clearly inferior to wOBA and wRC+. I get that OPS is easier to understand mathematically, but wOBA offers a clear and cogent explanation for WHY things are weighted the way they are while OPS just arbitrarily assigns walks, singles, doubles, triples, and home runs 1x weighting, 2x weighting, 3x weighting, 4x weighting and 5x weighting. They're similar enough stats that OPS doesn't suck, but I'd love to see people here switch to the w stats.

I'm not gonna speak for Doc, but one of the reasons I don't use some of the more advanced stats in everyday posts is because this isn't a math site or a predictive analytics site, it's a baseball site.
Anytime you need to use the phrase, "my MIT friend's understanding" to explain stats, you're gonna lose a lot of your audience.  If this was a physics model, then that would be important.  But since I don't believe we're properly capturing everything with our advanced stats anyway (like the overlapping nature of good defense decreasing its overall importance when amassed instead of additive offense, or the effect multiple black holes have on a lineup that destroys the effectiveness of that lineup) I don't feel like losing parts of my audience to have increasingly complex mathematical formula debates, all of which will then be redone when the new stat du jour comes out.
There are sites for that particular type of conversation, ones that do a wonderful job of parsing numbers, sifting those grains of sand for clarity.  Yours, for one, or Fangraphs, or any number of others.  I use them, and read them.  Now, Doc is a SABR guy, and Matt absolutely is.  Matt will crunch numbers with the best of them, and actually is, at this point.  I'm honestly not a numbers-first guy, even though my day job is in accounting and I understand the underpinnings.
But the point and purpose of this site is to have in depth, philosophical conversations about baseball, and so many of the stats that are used are baseline, look-em-up-anywhere stats. Again, not to speak for Doc, but I believe that was intentional, to not leave anyone behind while we talk about the game we love.  We're not trying to discuss the 53rd digit in pi, but rather whether pi is applicable at all to this situation. I want to ballpark the issue and then discuss it, not stamp it definitively and publish it in a peer-reviewed journal, so the extra 10% accuracy of a different stat doesn't matter so much to me if some portion of my audience doesn't follow me on the journey with those stats. I don't think the point of this site is in proofs, but rather in the search for a truth.  Maybe that's the liberal arts guy in me talking, and maybe this isn't true, but I feel like (and have received messages to the effect that) there are people who simply get lost if they can't find the number on the back of the baseball card, but their input in a discussion is invaluable. Should they get familiar with wRC+ and how it's a better encapsulation of contribution than just a slash line? I'm sure it wouldn't hurt.  And we talk pitch vectors and other things here that might get a little numbers-y.
But if the topic is broad, then why are we splitting hairs to thread them through the needle we lost in the haystack?  Get some of the hay out of the way first.  Then FIND the needle, and then we can talk about how to thread it.
SSI: looking for the right needle in a field of haystacks since 2005.
Threading em is your job, m'man. :) Interested to hear Doc's take on it, though. And if people want a more advanced use of modern baseball stats here, I'm not opposed. Nobody wants to be the Joe Morgan of the blogosphere, pooh-poohing "advanced stats" like OBP.
I guess I just don't see us being at that point of scratching in the dust and hooting at monoliths while spaceships are being built on other sites. Not yet anyway.
~G

I mean, wOBA is scaled to the same good/OK/bad values as OBP, so it should be pretty easy to understand whether or not a guy's wOBA is good or bad just based on your familiarity with that much simpler statistic. And wRC+ is scaled the same way as OPS+ (as a % of league average), so if you can read one you can really easily read the other. And wRC+ is carried on Fangraphs, where the leaderboards are way more useful than BBREF's anyways. All you have to do is say, we're gonna use wRC+ instead of OPS+ in posts now, they're the same thing only wRC+ is better, and then ta-da you've started to make more accurate points.
The thing is... look, I can see the argument in favor of OPS, sort of. It comes from OBP and SLG, which everyone is familiar with, so everyone knows how it's calculated. But to me, it's not enough to know HOW a statistic is calculated. If I'm going to trust a stat, I want to know WHY it's calculated the way it is. That's why I dislike batting average; there's no rational reason for some of the omissions it makes. wOBA has a legitimate reason to be calculated the way it is, i.e. the 24-state frame of thought and the run expectancies in each state. There's no way to know a good reason why OPS is calculated like it is, because there IS no good reason. You just add OBP and SLG, because it's easy and reasonably accurate.
I don't need my statistics to be easy to calculate. It's not like I'm the one calculating the OPS, or the wOBA, or the FIP... we get all this stuff off of leaderboards anyways. I want my statistics to be accurate, and if I'm going to trust that they're accurate I want the design process behind them to be rational. In that regard I trust wOBA far more than OPS, and wRC+ far more than OPS+.

It overemphasizes getting on base among other issues (ie, the way slugging is calculated) which makes it a very rough stat.
But my question to you is: are you having problems with the conclusions being drawn and illustrated with mathematical fingerpaint, or is it personal preference? Or, to put it another way: are we flat out WRONG about things that we say because we're not using wRC+, or is it simply less exact while stating what you would have stated more exactly with wOBA?
When I'm on blogs I get frustrated with the misuse of the English language.  All readers, for future reference: please do NOT use "myriad of" when talking about "a lot of" something.  You can use "a plethora of" or you can use "myriad." But "myriad of" is a no-no.
A lot of hits.
A plethora of hits.
Myriad hits.
That said, my pet peeves (it's not "per say" or "thier" or "suppose to") are mine.  You'll almost never see me correct anyone's grammar unless I simply cannot understand what they are trying to say, because it's the content of their message that matters.  We have several posters here from foreign countries, for whom English is a second, or third, or fourth (!) language. We actually have had drive-by posters drop grammar corrections on them, like there's a gold star for correct use of the word myriad instead of listening to the amazing insights they are making on a foreign board. And since anyone can make a thread here, the barrier for making a quality thread is not best-stat accuracy, it's thoughtful content.  No demerits for using OPS by any guest poster. Again, that probably comes from my belief that while baseball can be analyzed in many interesting and compelling ways using mathematical breakdowns of various numbers, it cannot be "solved."  So I'm not trying to solve it.
I'm sorry the use of OPS and slash lines is irksome.  I completely get it.  But unless you find the conclusions out of bounds with reality (in a "blue sky is green" sort of way) then I'm sorry, but I don't plan to stop.  I'll use xFIP if ERA for a pitcher is terribly inaccurate, but I'll note it and say why.  I use slash lines to denote what KIND of contribution is being made, rather than rolling it all up in one number that (while more accurate toward contributions than OPS for sure) can't be broken back out into its component pieces.
I do agree with you on wRC+ as a better "summary stat" than OPS+ but if I'm already using the slash line for ease of use and parsing, then swapping over to wRC+ is sort of like dropping standard for metric halfway through a conversation. *shrugs* I'm trying to stay consistent and tell a story, and I suppose as a writer the story is not made invalid by the use of first-level stats instead of second or third level ones.  Mileage may vary on that one.
----------------
When I was a kid, I loved astronomy.  One of my books talked about Saturn's density being the lowest of any planet, and that if you had a bucket big enough big ol' Saturn would float in water.  While not entirely accurate, it was illustrative. As this guy explains, he has a pet peeve about this "floating in water" statement. But 8 year old me thought this explanation about planet density was the cat's meow, and I used to look at Saturn through a telescope and imagine this floating, ringed planet like a giant inflatable pool toy.  I studied all its moons, checked out the Voyager pics...
Saturn captured my imagination.
And I guess I care more about that capturing of imagination and passion than I do about utter stat accuracy.  Heck, I round OPS off when I'm talking.  As long as when I talk about pi I keep it in the 3.1 to 3.2 range, I don't feel too badly about it (though my math-heavy friends do throw certain fruits and vegetables at me occasionally).  If pi = 5 in anything I post, please feel free to hit me with a wOBA hammer. I sincerely hope that's never been the case.
~G

I use slash lines in part because we talk a LOT of minor league baseball on this site, and there are limited 2nd and 3rd level stats for the minors.  Nobody's installed pitchF/X in the low minors, sadly. And I don't feel like we should be having one sort of conversation about prospects while having a completely separate conversation about the bigs.
Again, that's a personal preference not to relegate the minors to secondary status, and I enjoy flipping back and forth between the two conversations easily, so many of the stats are used with the intent to facilitate that.
I like your macro point about raising the accuracy of stat conversation, because we are no longer limited to stat conglomerations like OPS.  We can use a finer pen, get neater strokes.
But there are a lot of reasons we don't. I'm actually glad you brought it up, because sometimes the reasons that we do things become invalid, and we simply haven't bothered to update our processes since then.
Revisiting those decisions is always healthy.  If you see wOBA start to show up, that'll be your fault. ;)  But if it doesn't, I still heard you.  And I'm grateful you said something.
~G

Too often they display these real interesting looking stats, and completely fail to explain them in the glossary. And if they do, they neglect to inform us as to the abbreviation in their charts. Fangraphs frustrates me with this quite a bit.

Adrian Beltre's season wRC since 2004
161
90
106
110
106
81
140
134
141
138
Adrian Beltre was very consistent in Seattle, and he was very consistent when he left Seattle. Safeco was particularly toxic for him as a pull power righty, I feel that a fair weighting system would have less than a 30 point gap unless people really believe Beltre became a massively better hitter on his 30s after losing it from 25-29. The weights, to my knowledge, are not adjusted for handedness, and that seems like a huge hole to evaluating hitting talent. And what about quality of competition? The poor schmucks that have to bat for the Twins play in a division that has, as of this comment, 9 of the top 15 starting pitchers by FIP(Did anybody notice that Rick Porcello has panned out for the Tigers he's still the worst guy in the rotation, ranked 13th in the league by FIP), but I don't believe that would be represented in the weighting of wOBA. I would also complain that the easy the system treats all IBBs the same because intentionally walking Henry Blanco to get to the pitcher is different than intentionally walking Kyle Seager to get to Justin Smoak.
The point is, weighted systems don't take a bunch of factors into account, and a system that pays attention to some outside factors while ignoring others is nearly as likely to err as a system that doesn't pay attention to outside factors at all.

Um ... [tongue in cheek here] what bridge is going to collapse if we're wrong?  is the space shuttle going to explode?  Maybe Carlos Silva's family is richer than it "ought" to be, but other than that ...?  There are times when extra precision matters, but getting an extra degree of focus on baseball stats ... ?  Is there a downside that's worth arguing over?
 
Gordon's point about minor-league stats is obviously my point too.  I had to construct my own stuff because there just isn't the depth of data.
 
I think any stats that incorporate standardized results from BABIP but don't take into account the speed of the batted ball have built-in problems.  [In my stuff, I just generalize to the distinction between "balls hit with authority" and "random-y balls in play."  But the idea is not every batted ball has the same BABIP.]  And that data probably will never be available for minor-league parks, or at least not anytime soon.
 
I dislike all the stats that try to put all of baseball into one number.  Even OPS+ ... I prefer to break it down to the two component parts and the composite (the "three numbers").  OBP is yes/no while SLG is weighted. So two different things. Out/not out vs. gaining bases.  I don't even like smushing those to things together.  Trying to add defense too?  and base running?  All in one number (like WAR)?  It's helpful, don't get me wrong, and it's fun to see how it comes out, but to act as if it is a "real" number resulting from precision engineering, that has some sort of "gold standard" value is just too much to ask in my opinion.
 
I said in my very first posts about the "Spectometer" stats, what good is a stat if it alienates the regular fan from the game?  If the reader has to know more Stephen Hawking than Stephen Pryor to get through a baseball blog then what are you communicating and whom are you communicating it to?
 
Give it time.  "Regular" fans are getting to know OBP and OPS.  Mainstream media are using them.  BB/9 and K/9 are increasingly common instead of just W-L and ERA.  The stuff that actually conveys meaningful information will eventually seep through.  "The folks" will get it if it makes sense and communicates something.  When I first started working for my senior partner (now 71) people laughed heartily at the idea that he would ever sit down and compose an email.  Now it's probably 75% of his day.  Times change.
Well, that didn't end up being so quick, did it?