Hacker News Reader

Comment by mattmaroon | Hacker News Reader

As usual, Gladwell is cherrypicking data to display something unrelated to reality. Let's look at the first QB picked since 1999.

2000: Chad Pennington, this year starting qb for the Jets, doing well.

2001: Michael Vick, who was one of the top few QBs in the league. Of course talent scouts can't predict animal cruelty.

2002: David Carr: Playing backup to Eli Manning. Started for 6 years until an injury sidelined him. Not a McNabb, but far from a Tim Couch

2003: Carson Palmer: 2 pro bowls with the Bungles.

2004: Eli Manning: Superbowl MVP

(here's where players are still too new to tell much about)

2005: Alex Smith: that one hasn't worked out yet and may never.

2006: Vince Young: Promising backup on the Titans.

2007: JaMarcus Russell: Starting for the Raiders

2008: Starting for the Falcons.

Either way, in any non-Gladwellian universe, I'd say college QB scouts are doing pretty well.

View on HN

Hmmm, who's cherrypicking data? ;)

2000: Chad Pennington, Pick 18, Career Rating: 89.7; Marc Bulger, pick 168, Career Rating: 85.7, Tom Brady, Pick 199, Career Rating: 92.9

2001: Mike Vick, Pick 1, Career Rating: 75.7; Drew Brees, Pick 32, Career Rating: 89.1

2002: David Carr, Pick 1, Career Rating: 74.6; David Garrard, Pick 108, Career Rating: 85.4

2003: Carson Palmer, Pick 1, Career Rating: 88.9; Tony Romo, undrafted, Career Rating: 96.6

2004: Eli Manning, Pick 1, Career Rating: 76.4; Big Ben, Pick 11, Career Rating: 89.9; JP Losman, Pick 22, Career Rating: 76.9

2005: Alex Smith - Noted; Aaron Rodgers, Pick 24, Career Rating, 89.8; Jason Campbell, Pick 25, Career Rating: 80.5; Matt Cassell, Pick 230, Career Rating 84.2

2006: Vince Young, Pick 3, Career Rating: 68.6; Jay Cutler, Pick 11, Career Rating: 88.6

2007: JaMarcus Russell, Pick 1, Career Rating: 67.5 - uhhh, "top" QB

2008: Matt Ryan, Pick 3, Career Rating: 92.0 - very good - so far

So, of the last nine years, the top QB taken has turned out to be the best QB of that year, maybe once. Russell just isn't good. He won't be starting for long. And Ryan, it's too soon to really tell. Meanwhile, QBs taken, sometimes very, later have often been as good, if not better, than the top QB's in 6 of the last nine drafts you cite. There are just as many guys like Brady, Bulger, Garrard, Romo, and Cassell (picked well behind "names") as there are top picks that have panned out.

EDIT: Forgot Romo.

You might as well add Flacco to 2008 too. He's doing fairly well starting with the Ravens.

-- Delaware alumni

I didn't cherrypick, I simply pointed out that the QBs scouts rate the best generally perform very well. Every now and then a Tom Brady or Kurt Warner sneaks up out of nowhere, but for the most part almost every starting QB was first round drafted and often first overall.

Also, don't confuse best passer rating for best quarterback. All sorts of things factor into the QB's passer rating, such as the quality of his receivers and offensive line, his team's style, etc. Those numbers aren't very meaningful.

Nobody would ever suggest that scouts could get it perfectly, or near perfectly, in order, but they do a pretty fantastic job. You can cherrypick data to show a few black swans, but for the most part they're highly accurate.

But it's not "every now and then". Look at the names of "nobodies":

Brady, Bulger, Garrard, Romo, and Cassell

And that's just the last nine years. They're as good, if not better, than "top" picks. Then you have the more historical guys like Wagner, Young, and Montana.

Picking for the QB position really does seem stochastic.

QB is not stochastic at all. These are all quarterbacks who arrived at a team that was already good. "Quarterback rating" really measures the goodness of the whole passing offense. It is no surprise that the Patriots, Rams (before this year at least), Jaguars, Cowboys, and Patriots again can give an average QB great stats.

Right, it's a terrible metric for determining the effectiveness of the individual.

Now, you're arguing with yourself. If it's the "system" that's getting "rated", then there's even less reason to assume a team knows what they're doing when they're picking a specific QB.

And that's the interesting bit of logical jiu-jitsu going on here. The teams picking first are the ones that are usually the worst to begin with. Selecting higher just doesn't help them to improve their lot when it comes to the QB position.

You are ignoring the huge difference between "teams are choosing the optimal QB" and "teams make decisions no better than random". The decision is neither perfect nor random.

This is getting far afield of the original critique. The point there was that scouts do a good job in selecting QBs, especially for picking the best QB (first!) in a given draft. There is very little evidence to support that claim.

Surely, selecting the top college QBs in the country isn't random. But selecting among that echelon appears to be, especially for the transition to the NFL game.

I think you misunderstand, because I'm certainly not arguing with me. I'm arguing with your methodology. Scouts do a solid job of picking the QBs that are the best.

Passer rating does not measure QBs skill accurately, therefore scouts may not pick the QBs with the best passer rating. That's because their job is to pick the best quarterback, not the one who has the highest number in some goofy equation that takes many other things into account.

QB rating is bunk. Chad Pennington has a better career rating than Dan Marino. Dan Marino is an all time great, Pennington might just barely crack the top 10 current starters.

So there's a correlation with where a QB was picked and their performance? But you can't tell me how to, objectively, judge their performance? What, we're back to relying on the teams and how they dole out playing time?

I see little evidence suggesting that where a QB was picked correlates with their skill at the position. You named the last top QBs picked in each draft year and gave the most positive assessment possible of them. I pointed out that for almost all of those years there were better QBs drafted below, and many times far below, that consensus best choice. That doesn't give me great confidence in the scouting of the position. You can't say teams do a "a solid job of picking the QBs that are the best" if every year they tend to miss the best QB available, and oftentimes by wide margins.

Now we're arguing about the role of a "system" in producing performance numbers for a QB. Again, if that's the case, then the individual QB means even less than your original assessment, where the highest weight was placed on the highest picks. If anything, there's even more reason to distrust the team's picking highest (and their scouts) because they wouldn't be in that position if their system was better. If you say a team's system, already in place, will predict a QB's future numbers (your critique of passer rating), then you're also devaluing the importance of a high pick.

The real problem is that teams weight the position too highly AND create a dearth of talent based on their selection process and the limited skills/pedigree they value. That leads to a Wal-Mart stampede in the first round, doling out big bucks, to players that won't a change a team's fortunes one way or another. The position is too dependent on many other positions.

I see you're from Browns territory. How many swings have they had at the "best" QBs in a given draft and how has that worked out for them? If anything, they've continued to make the mistake of valuing the position too highly to the detriment of the rest of the team.

Funny though that you're touting Marino's greatness. He slipped because scouts devalued his "head" and his greatness is only indicated by his raw numbers. A better example for your case, would be someone like Bradshaw or Aikman. Both were the top overall picks in their drafts and both took poor teams and eventually led them to dynasties. Then we have guys like Brady and Montana...and we're back to where we started.

Looking at your more complete list, I see a bunch of QBs who were first picked who did well, with only one or two exceptions. I see some QBs who were picked second or third who also did well. (Overall picks aren't really fair, as teams have many reasons for selecting other positions.) And I see a couple black swans, like Romo.

That's what you would expect if they were really good at their job. If they were bad at it, you'd expect more first picked QBs sucking, and more people coming out of nowhere.

Objectively measuring is impossible. Passer rating is more an indication of the quality of a team's offensive line, receivers, etc. than QB skill. It's a nice number for commentators to throw out, but it's of little use to a statistician.

I'd say maybe Pro Bowls, but again, there are other factors in play there too.

So, of the dozens of qbs drafted every year, 4 or 5 surprises popped up in a decade and that makes QB selection stochastic?

No, the fact that by the time they're eligible for the draft, and of the 10-15 taken every year, there's usually little to differentiate the top picks from the worst. What separates them is anyone's guess. That's Gladwell's point.

But if you look at the top picks and the bottom picks, you find almost none of the bottom picks starting in the nfl (you named a few who did it in the last decade) and a ton of the top picks who are, or are at least second strings to superstars. I would say that means that what separates them is far from anyone's guess. There's certainly some variance there (see Tim Couch) but it's not unpredictable.

The complicating factor, of course, is all those former first rounders but second stringers, have had significant money invested in them. It's a lot easier to cut the lower picked players but not necessarily because of their "talent".

Matt Cassell = perfect example.

Does he even get a chance on any other team? Ever?

A lot of great QBs got their starts when the superstar they were behind got injured. Steve Young for instance.

The picks from 1990 to 1999:

1990 - jeff george - fair? 1991 - dan mcguire - BAD 1992 - david klingler - BAD 1993 - bledsoe - GOOD 1994 - shuler - BAD 1995 - MCNAIR - GOOD 1996 - Tony Banks - BAD 1997 - Jim Druckenmiller - BAD 1998 - Manning - GOOD 1999 - Tim Couch - BAD

At best I would say the #1 qb pick has been 50/50 historically.

You are cherrypicking even worse. Just because the highest rated QB was not the top QB does not mean the top pick is not the best. You should at least consider all the terrible quarterbacks picked in the later rounds.

Plus, the top pick always goes to the worst team in the league the previous year. I am more impressed that Carson Palmer has managed an 88.9 rating with the Bengals than I am than that Big Ben has managed an 89.9 with the Steelers.

There are 10-15 QBs taken every year out of 250 picks. The question is whether where they were picked correlates with future success. I see no evidence to support that correlation.

Otherwise, we're taking about a very limited skill set (i.e. it translates to nowhere else on the field). The reason so many QBs are picked high isn't because their talent deserves it. It's because teams overvalue the position AND there are few QBs deemed worthy enough to even be considered.

New England is actually an interesting case study of what it means to take a QB who may not have the typical skillset, but if you place them in the right environment, they can be a true star. The point is: Their success argues strongly against the role of a high pick for the position.

Plus, the top pick always goes to the worst team in the league the previous year.

This is key. Growing up in Tampa, I watched a lot of high draft picks have great careers after they left the hell-hole they got drafted into. I've always felt that getting picked as a QB in the late first round would be best, because you could be on a good team, rather than start right away and get pummeled in the 0.003 seconds you have in the pocket.

Gotta correct you on Young. He was a promising starter who many, including myself, thought could make the Pro Bowl very soon (especially if he played with wide receivers). Then he became a head case and lost his job to a mediocre Kerry Collins.

Also, I think you should include the second and third quarterbacks taken to round out your stats...that might be interesting, with stars like Ryan Leaf :^)

A better test of if QB scouts are doing well is if the #1 QB pick turns out to be the best QB in the draft.

In most of those years you listed, there was a better qb left on the table after the first qb pick.

Out of the 9 years listed it really only happened 3 times that a better QB was left - 2000 (Tom Brady), 2002 (David Garrard - debatable, but he's starting and Carr isn't), and 2005 (Aaron Rodgers). So 6 times out of 9 the best one was taken first - that seems pretty good.

If you go back even further I'd imagine this trend would continue - even Gladwell's main example (Ryan Leaf) was taken second to the best QB playing today (Manning).

Yeah, Garrard/Carr is debatable. Garrard wouldn't be starting over Eli Manning either. So it may be 7/9.

not to turn this into sports news but:

2003: romo 2004: roethlisburger 2006: cutler 2007/2008: way too early

Historically speaking: steve young, montana, marino, and I could go on were not thought of highly by scouts

I thought it was commonly accepted that scouting QBs is a guessing game.

I'll give you 2003, not 2004, and I'd lump 2006 into too early to say for sure though you're probably right.

Yea, I don't want to get into a huge debate about this either, and the names you mentioned are definitely in the same class as the first picks. My main thought was that while its not a science, the scouts are definitely not just randomly picking players like Gladwell makes it out to be (saying that "there is no way to know who will succeed at it and who won’t" and that "college performance doesn’t tell us anything").

The rhetoric in the article is a little ridiculous, but for a moment let's gloss over his glossed over details:

Professional football teams have an extreme financial incentive to accurately predict an athlete's potential. They have a sizable data set to examine, and a lot of money to spend, but there have still been some very notable failures.

Our education system does not have the resources of the NFL to determine who will be a good teacher. But even if they did, it would be a misappropriation of funds. Instead of focusing on pre-facto credentials (graduate degrees, etc.), we should put them in the mix and see how they perform, and pay the best teachers accordingly.

These seem like the most salient points from the article, and I don't think any of them are particularly false or oversimplified. If you know a lot about something and the article is for a general audience, it's bound to seem like a bunch of outright lies. Life is never as elegant as the New Yorker makes it seem, but that doesn't mean it's not entertaining (and sometimes informative) to read.

"Our education system does not have the resources of the NFL to determine who will be a good teacher. But even if they did, it would be a misappropriation of funds. Instead of focusing on pre-facto credentials (graduate degrees, etc.), we should put them in the mix and see how they perform, and pay the best teachers accordingly."

Thanks for trying to get us back on point ;)

What's also interesting is how having the top pick cripples the drafting team due to the ridiculous salaries given to unproven talent. If I had the top pick, I'd trade it every time. 50/50 chance isn't good enough for allocating 8 figure salaries. The NFL needs a rookie scale like basketball.

Bungles. I see what you did there... ;)

I'm from Browns territory, sorry.

Oh I'm not a Bengals fan - just never heard that clever team name before. Well played!

"just never heard that clever team name before"

You obviously haven't been watching ESPN for, oh, the past decade.

Ha! I'm a programmer, what can I say? ;)

Yeah I don't think ESPN.com is getting many pageviews from this crowd.

Reply on news.ycombinator.com