Highly relevant to your comment is episode 6 of Malcomb Gladwell's Revisionist History Podcast [0]. In the episode, Gladwell asks the president of Stanford whether there was any amount of money he would turn down because he thought it would better spent somewhere else. He replied, somewhat evasively, that no, there wasn't really an sum large enough. It's a truly incredible interview to listen to.

I highly recommend the three part series about college education that makes up episodes 4,5,6 of the season 1. But, if you have the time, I think the entire podcast is excellent while noting that I am not a huge fan of Gladwell's books.

[0] http://revisionisthistory.com/episodes/06-my-little-hundred-...

> He replied, somewhat evasively, that no, there wasn't really an sum large enough.

Which, of course, is his job. I wouldn't be surprised at all to find that he has an entirely separate point of view in private when being candid.

That said, my immediate reaction to reading the first sentence of the GP comment was to think of that same Revisionist History episode. I'm excited that the new season is beginning soon. I will temper that with a caveat that I think Malcom does push his agenda and narrative fairly hard in those podcasts, but that does make for a really compelling story. If you can separate the facts from the portions it feels like he's omitting, you'll likely get more out of the podcasts, but I find the exposure to new ideas particularly fun and useful. I especially liked the Moral Licensing episode, which I only listened to for the first time a few months ago. It's particularly fun to listen to that one and realize that his specifics were all wrong, but the generalities and basics still match perfectly with the new Political reality that was only 6-12 months int the future from that recording (regardless of whether you believe in the premise).

I'm being particularly vague on purpose here, because I truly believe that episode was at least twice as as entertaining and memorable because of that situation, and I don't want to ruin it for anyone. That episode was so good, it's probably worth a re-listen now if you listened to them as they were released.

> asks the president of Stanford

Interesting tidbit: the president of Stanford at the time was John Hennessy. Take any modern CPU, and you'll probably find that much of its internal architecture is built on work by Hennessy and his group at Stanford. He is arguably the greatest computer architect of all time.

He's definitely a powerhouse, but I think Patterson is a stronger contender for that title.

They would definitely both be among the very top candidates to receive such a title.

Any time someone brings up Stanford, I want to scream patent troll. If you sell to a patent troll, you are a patent troll.

http://news.stanford.edu/2015/02/23/haber-patent-trolls-0223...

Shame on you, Stanford! What a sorry college.

Fortunately they collaborated and wrote a book on the subject!

*THE book

:)

I don't have much to add to this beyond another strong recommendation that you listen to that 3-part series; it's the strongest work he's ever done in any medium.

I just remember 2 episodes. While it was entertaining, one episode lacked any intellectual rigor at all. It seems like Gladwell can't understand the counterarguments at all and chooses to cut out most of them with out realizing.

One obvious thing he misses is the moral hazard of rewarding poor performing schools. Also, the real question is why donate to universities at all? There is no real reason why universities can't support teaching purely on tuition. University can easily charge on a sliding scale based on need and SAT scores. If they want more kids to go to college, they can give to scholarships directly. Giving to university directly just promotes spending them on outspending each other on other axis for attention...

The "obvious" point you're making in your second graf is one Gladwell agrees with.

Is the second reason one he actually agrees with? Does he consider the second order effects? He thinks giving more to poor schools would be a "good" thing. All it would do is cause more schools to try to fight for prestige and donations. We have 15ish schools were their number one goal is to find ways to get donations. What would happen if 150ish school expected to be able to get huge sums of money as donations? Many would change their focus away from education and what students what to donors. I think it likely to increase tuition and decrease enrollment. Top 15 schools need to be selective to be the best. If the 100th best school want to look like it getting better it needs to become much more selective.

"I'm not convinced that this outweighs the benefits of funding hundreds, or even thousands of students/scientists in less prominent schools who might go on to make such discoveries themselves"

While it's plausible anyone can discover anything, the only rule of thumb about finding great research is that great researchers generally train great researchers - or, to put another way, it's likeliest to end up doing nobel price worthy research by being trained by a nobel price winner.

The other thing is bad schools do not turn magically into good schools by pooling money into them. Improving education requires human intervention by experts - experienced teachers and so on. Money helps, but if you are sitting on a 140 million dollars, you can't just donate it to some struggling place and imagine it will transform it to a better institution. Or, pick random ten laboratories that have done unremarkable research and imagine a grant of a 10 million dollars will improve the quality of their output.

I think the only rational choices to well spend such a sum of money is to support established high performers, or, to proactively invest the money. If proactive involvement is not what the donor wanted then I can't really fault his decision to go with MIT. Or, well, if we follow the silly "nobel price winners beget nobel price winners" rule of thumb and use nobel prices as metric of research quality which itself is also silly, there would have been other options: https://en.wikipedia.org/wiki/List_of_Nobel_laureates_by_uni...

There's a flaw in this reasoning: are Nobel winners actually training people better or do they simply have access to more money?

In my experience researchers that get a lot of grants are able to fund lots of projects, which leads to smart people working for you, which leads to big publications, which leads to more grants. Nobel/Lasker/Name-your-prize winners are just some of the people at the top of the grant pyramid. In fact, I'd argue that who wins a Nobel is a bit of a stochastic/political process, there are tons of scientists who made Nobel-worthy contributions that never won one.

That said, the outcome ends up being the same. Big/well funded labs tend to be started by smart people who had the advantages at big/well funded labs and the cycle continues, but I think the way things are funded is a big part of how that system works, not the inherent smarts/brilliance of the Professor.

Also it's not about pooling money. I think the OP is making a comment on effectiveness for educating people. For example, Cooper Union, an exceptional institution in NY that has a history of training people to an MIT-level for essentially free has come upon hard times and has somewhat abandoned that free-education principle (granted this was due to some financial mismanagement, but a big donor could certainly force changes there). This was an institution that taught many disadvantaged students, particularly from the NY area who couldn't afford MIT and its compatriots. Perhaps the money would have had far more impact there? Or what about offering massive scholarships to people attending community colleges who have a helluva GPA but need help for childcare?

$140M is an incredible gift, I can't help but think there are better ways to spend it than giving it to an institute that already gets massive amounts of grant and other money. Maybe a place that could really use it to help those that need it most?

I merely pointed out one plausible statistical strategy for distributing a donation if the idea is to cause an impact and not distribute the funds back to the economy more or less randomly - I have no idea what the underlying mechanism is.

"Or what about offering massive scholarships to people attending community colleges who have a helluva GPA but need help for childcare?"

That's proactive. But, then again, we know nothing on this donor.

"Maybe a place that could really use it to help those that need it most?"

And what place would that be? Who would exactly be those that 'need help most'? Where I'm from government tries to handle the material well being of those who've had the shortest end of the stick in life - no idea what charities would provide the best bang for buck in US.

I would not call spending money on science wasteful in any case.

Two different issues- research labs/graduate training quality vs undergrad lectures and such.

I doubt MIT teachers of basic science are the best in the world. Rather MIT attracts some of the smartest most motivated students in the world. More money won't do much for this.

Research lab funding? I think at it's current steady state more money sent to MIT for research would likely find a useful currently unfunded study or group.

Mark Zuckerberg generously gave 100M to improve the education system of Camden New Jersey. At the time the mayor was Stanford grad Cory Booker, now Senator, and future Presidential material. Camden was and still is a minority shithole of New Jersey. Neither Mark's generous gift nor Cory's public service had a lasting effect. Deeper structural problems. I do not know the answer.

Except the gift was nothing but funding the charter school grift (don't get me wrong, charter schools could work, and in some cases do work, but for the large part, in the US, the charter school movement has pretty much entirely been hijacked by union busters and people who want to profit off the privatization of education).

Exactly. There are plenty of laboratory schools attached to universities, where teaching research is done under the eyes of the public. Those are as a rule doing well but Zuckerberg chose to boost unaccountable charter schools. Glad to see that it went as one would have predicted it to go.

It was Newark, and the half donation went to Charter Schools and consultants. [1]

1. http://www.businessinsider.com/mark-zuckerbergs-failed-100-m...

"Minority shithole" is a bit harsh... I spend a lot of time in Newark and it is a much better place than it was twenty years ago.

The Ironbound is getting nicer every year. Yesterday I was walking home from the train and there were young people everywhere. Definitely not what I am used to around here.

Granted, the Ironbound could be a totally different city than the rest of Newark.

It was actually Newark...

Zuckerberg gave to Newark, not Camden, but you're right in that the donation didn't have the intended effects.

On the other hand, more money means you can afford to pay for higher quality professors/staff/technology.

I took a "big data" course on Map/Reduce, Hadoop, etc. We had to use VMs on our own laptops because the department couldn't afford to get us temporary AWS instances or similar.

Sure money is not a quick fix for a bad school, but it opens up more opportunities than not.

Good salaries are necessary, but not sufficient to attract top level talent. If the whole environment is mediocre, it won't be able to produce better research or education, regardless of money.

(Speaking as someone who graduated #1 university in Poland, but nowhere near #200 globally...)

Regarding your 2nd point, do those rankings actually provide any useful information? How is it fair to rate an entire university, especially when they offer multiple studies? Shouldn't the ranking be based on discipline e.g. software development, language, *ologies, etc.?

Also #200 out of what? According to this Quora[0] question, there are anywhere between 26,000 and 40,000 universities globally. 200 out of 26,000 puts your university in the top 1%.

https://www.quora.com/How-many-universities-are-there-in-the...

This is actually part of the strength of a large gift to a university like MIT. Using the prestige of the university, and the money, you can bring in an entire cohort of capable scientists dedicated to a research area, and achieve scale via a "brain trust" that would otherwise be impossible. Other universities have indeed done this in the past with great success.

I like how the accepted dichotomy in this entire debate here in the comment is either MIT or bad school. There is an in between! There are lots of competent, well managed, and important educational institutions outside of the top 2-3 schools! Maybe once you get to like 30-40 or so, maybe that argument holds more water. But let's not pretend like plenty of state universities aren't great academic institutions.

I think the best educational utility of those millions of dollars would likely be subsidizing tuition for two year schools.

In case you're curious, this has actually been studied recently. Researchers compared funding in US and Canada and found that

- science impact is a decelerating function of funding amounts

- science impact plateaus around $700k for a lab or project

- ancillary impact benefits include more diversity by funding more groups with smaller amounts instead of fewer groups with larger amounts

In all of this, science impact is measured in publication rates, citation rates, and most cited articles over a four year period.

http://journals.plos.org/plosone/article?id=10.1371/journal....

Which is a kind of stupid way of measuring things, the detector systems at the LHC were developed by a relatively small number of groups at huge expense over at least 12 years, it took even longer until actual science experiments happened. A concrete example would be the Atlas trigger systems development until Higgs discovery was ~20 years, ultimately they received >50000 citations.

Anecdotal. A big grant paying off doesn't outweigh all the other countless iterative and great leaps forward that led to the detector which led to the higgs. Those efforts were also funded by small grants, graduate student stipends, or out-of-pocket in pubs and libraries.

The reason why they measured impact in that way is because that's what most funding agencies today are speculating in for their decision making processes. And as data shows, most funding agencies are getting it wrong.

Pretty much all large scale physics projects have payoff profiles like that and that is probably known to funding agencies, so I severely doubt that funding agencies have truly used this as their measure. That way to measure impact is simply wrong and would have prevented many significant discoveries in high energy physics and astronomy.

ITER, Wendelstein-X, LIGO, Planck, I could go on. In all those cases any results happened >10 years after the funding started or might still only happen in the future. In the case of the LHC, but also many other Detector experiments individual grants to research groups were typically (not just anecdotally) way beyond 700000€.

Are you claiming the impact/$ for the LHC is higher than would be expected from funding many smaller projects?

According to wikipedia the project costs about $9B, but I doubt that includes all supporting efforts and researcher salaries. That would fund 12 thousand of these smaller $700k projects.

I think "impact" as typically defined is a questionable measure of scientific value. There is no other way in which the Higgs could have been discovered except through a concerted effort of thousands of people, the technical and infrastructure work in such a project is both very challenging and expensive and usually pays only off after more than one decade of work.

You are completely missing the point. The detection of the Higgs is not the primary and absolute purpose of humanity. That it takes a massive amount of money and effort is a bad thing and it certainly isn't something we needed to do right now.

If we first invested in research to make building the LHC cheaper, we'd get more benefit for less cost. That's the only reasonable criteria for allocation of resources.

For some reason this reminds me of how hard it is to design unsupervised learning systems that don't hit a measurable payoff for a long time, so they can't easily decide what they did right or wrong.

Also, I love your username :)

Is MIT actually well-endowed? I don't know the answer to this question so it's not intended to be snarky.

But there are plenty of schools (my alma mater CMU is one such example) that have mind share but a very small endowment. CMU has made a point of building its endowment over the last few years -- there was a single $250M donation, a few others in the $50-100M range, and calls for much more.

Endowments are important to the long-term financial health of an educational system. They help schools borrow money at very cheap rates to fund large infrastructure projects that otherwise would need to be built by donor / grant money alone. A school like CMU with a small endowment ends up having sky-high tuition to compensate (indeed, CMU has the highest tuition in the country across nearly every degree program they offer) and limited scholarship opportunities. They also have to accept a greater percentage of foreign nationals who otherwise might not qualify (and who can usually pay the tuition out of pocket no problem, since the wealthy across the world want at least one degree from a US school on their resume).

But essentially, the endowment is a nest egg that the school uses to ensure its long-term survival and academic independence. If you feel your alma mater made you the person you are today, and you want to encourage that to continue into the future, a large endowment gift goes a long way to ensuring its continued existence, independence and prosperity. Agree that there are a few schools that don't really need more (Harvard being the prime example) but the vast majority of schools really do need that much money sitting in the bank. You're donating to a school of thought more than anything else.

That's not really difficult to find out. MIT's endowment is the sixth highest in the US. In 2016, it stood at just over $13 billion.

MIT is known for having a good financial aid program and a large endowment (MIT's is around a third of Harvard's!) could improve that even further. I can imagine that being very impactful in the long term.

72% of MIT undergraduates complete their degree with no student debt [1] and the rest are from families in the top 1-5% of income [2]. Thus, if directed at financial aid, this pledge would only help those from high-income families. If the pledge were used to give a stipend, could MIT attract more low-income students to apply and accept their offers?

Giving students from high income families a tuition discount likely won't change their decision to attend, hence price discrimination. To maximize charitable impact with respect to financial aid, it could be argued that instead the funding should go toward other schools so that these all-star, low-income undergrads have more opportunities closer to home.

[1] http://web.mit.edu/facts/tuition.html

[2] https://www.nytimes.com/interactive/projects/college-mobilit...

What if the money was used to expand and take more students? (rather than subsidising the ones they already have)

I think this would work, but university prestige is often based on scarcity so few universities do. This is the same reason that Harvard has almost the same number of undergrads as they did 50 years ago, despite an endowment 20x the size.

> university prestige is often based on scarcity

But it is a combination of scarcity and quantity, interestingly.

By any measure that is per-student, Caltech grads for instance are far superior to the average MIT grad.

However, MIT has better name recognition and looks far better in stats like "number of publications in top journals" or "number of Nobel price winning alumni", just because it has five times as many researchers and pumps out five times as many alums every year. So the winning strategy is to get as big as possible - but only as long as the majority of your intake is still in the top half-percent of the bell curve...

> By any measure that is per-student

That's a pretty broad statement. I can think of a number of measurements where MIT comes out ahead of Caltech - entrepreneurship (measured per-student) would be one particularly relevant to HN.

Is this really true? I always thought this was due to the ridiculous real estate prices in Cambridge.

Cambridge is certainly no where near as expensive as say, New York, so by your logic Columbia should be smaller than it is? A significant portion of a school's ranking (and overall prestige) is the difficulty of admission and the yield, so schools have an incentive not to expend indefinitely (plus there are other advantages of small/medium size).

I personally know several people who were not in that 1-5% who did go to MIT and most definitely have massive student loans. Maybe they do that now, but they definitely didn't in the past.

More acendata: In 2009, I forewent MIT because it would have cost me ~$40K a year whereas every other Ivy I got into asked for $15K to $20K a year.

Agreed, there could be people outside of the top 5% who have loan payments, and those from families earning more than $200,000 who decided not to pay the full amount.

With this pledge, maybe MIT can bump up the free tuition level for those from families earning under $80,000 to $125,000 like at Stanford. The number is somewhat arbitrary and depends on the sliding window of tuition rate. After all for well-endowed, prestigious schools, tuition is unhinged from services received / amount of spending per student.

That would mean that the people you know are in the remaining 18% of people. It does not contradict the parent post.

Financial aid packages often assume a parental contribution before loans and grants are considered. When there is none, it's on the student to cough up the difference or withdraw. There might be more room to help prospective students than you think. (Speaking from personal experience.)

Hmm, don't think I understand your first argument. If 72% of MIT undergraduates graduate without loan debt, this almost certainly includes all of the students (29%) coming from the top 5%.

There are certainly people that are dissuaded from MIT for cost reasons, and money could be used to solve that problem.

No, that doesn't follow. Students coming from the top 5% can have loan debt. The top 5% includes many (mostly?) families that could not afford to pay out of pocket for MIT, and even those that could may choose not to.

Financial aid is just rebranded price discrimination.

That is literally the term for providing the same product or service at different prices to different markets depending on their perceived ability to pay, yes.

You seem to be implying that it's necessarily a bad thing, though.

>where the money might arguably be more effectively used or sorely needed.

Malcolm Gladwell did a great podcast episode on this very topic: http://revisionisthistory.com/episodes/06-my-little-hundred-...

Highly recommended.

It seems like 140 million would be enough to create something along the lines of a MIT OCW or Khan Academy-with-testing-centres to earn degrees.

Maybe it's the unrealistic and the naive opimist in me that is saying this--but imagine any person in the world with an internet connection could have access to study for a world-class degree on whatever schedule suits them?

Obviously there are challenges involved, and there will always be a place for research institutions, but it seems to me like if anyone in the world could earn a degree equivalent in rigor to an MIT degree that a lot more would. That would help a lot of people and also propel advancements science and tech.

It seems most of us in tech learned a lot of our most valuable skills on our own--so in other words, labour-intensive education isn't the only way to do it.

But...it's not the degree, it's the knowledge that's important.

MIT, Stanford, and a decent amount of other institutions have made their lectures available for free online. Almost anyone can go online now and watch the lectures and do the projects that the students do.

How would testing centers and degrees make them better scientists in anyway?

Because of limited resources. A degree signals knowledge of a subject, ability to commit and see-through to a goal, and of intelligence.

It also decreases uncertainty in the outcome of your time investment.

>How would testing centers and degrees make them better scientists in anyway?

I'm not aware it's even possible to get accepted into a phD program in the sciences without aN undergrad degree.

Utility is necessarily subjective. Presumably the anonymous donor decided the best way to maximize their utility was giving it to MIT.

State Universities are getting underfunded these days. I would like to see a major challenge grant for state universities with major science/engineering schools. For example, donate $140 million if the State increases its expenditures by $140 million above current trends for each of the next 5 years.

They also don't typically operate government labs, which is a good way to pay for infrastructure and facilities and toys.

I've thought about this a lot relating to charitable giving, especially because my employer frequently offers 2x match on these kinds of donations. How does one compute 'bang for the buck' for a given donation size? How does one balance small, local socially-good institutions vs large, global socially-good institutions?

Also, some global institutions do small scale local outreach as well. Despite my employer being a large global research university, we have health clinics in low income areas and provide educational opportunities for low income neighborhoods locally.

Or a single donation of that size to another single school that is far more in need than MIT. for example on could donate it to the Tribal Colleges system and make an unbelievable impact on the lives of many criminally oppressed native communities.

There is low level good and high level invent nuclear fusion kinds of good.

This is the type of thing I think a lot about. My hometown is plagued by an outrageous crime problem, widespread poverty, and crumbling infrastructure. They just spent $250 million to extend a park over a highway, another $51 million on a trolley for tourists, $24 million to redo a downtown park, and so on and so on.

Obviously, the funds for these projects weren't fungible, but taken as a whole, this spending in the face of real problems is profane.

Whoever made the donation was clearly of an opinion that MIT, being a bunch of smart people can allocate the money better than the donor or other institutions.

>It would be interesting to compare the utility of one large $100 million+ grant to an institution like MIT, Yale or Harvard that is already well-endowed, compared to a hundred different million dollar grants to institutions across the USA or the world, where the money might arguably be more effectively used or sorely needed.

For example, UMass Boston.

I had this as the essay question on a standardized test (ACT). A ...loaded... question no doubt.

in this case, since it was done anonymous,I believe the impact and affinity is the prime motive.

rather a firm belief on part of the donor that the only way for his huge sum of money to do any good to this world, is to go to MIT.