Via Alex Tabarrok, here's the racial and ethnic breakdown of inventors in the United States. The figures are from a new demographic study done by Ufuk Akcigit and Nathan Goldschlag:
Every nonwhite group is massively underrepresented except for Asians, who are massively overrepresented. Women are also enormously underrepresented:
In this study, "inventor" is anyone with a patent to their name. It's almost exclusively a white and Asian male domain right now.
The conclusions of these data are obvious:
White men are much smarter than blacks, Hispanics, and native Americans. Asian men are much smarter than white men, and double-much smarter than everybody else. Also, men are much smarter than women.
Just goes to show ya, doesn't it? Obvious conclusions sometimes require further investigation.
The very fact that this is the immediate conclusion you leapt to shows how innumerates shouldn't be allowed to handle data...
Regardless of supposed power inequality issues, as a purely mathematical statement what you're saying is not the relevant deduction. What the data tells us is NOTHING about "white men" as a whole; all it tells us is that the fraction of white men who are better than average at "inventing" which sure let's use as a proxy for smartness, is higher.
In other words -- there are proportionally more smart white men, but that tells us something about the top 1%, it says nothing about the other 99%.
ONE way in which this could play out is having identical means for all these groups, but higher variance in white men, and even higher variance in asian men. Higher variance of course mean more at both extremes, better and worse. I don't know what the data show but as regards men vs women, this seems to fit ones intuition; there are in fact substantially more men than women intellectually disabled. (The numbers vary and can switch to more women, for other types of mental disabilities.)
But equal means and higher variance is of course only the simplest way this could play out. You could have equal means and equal variance but higher kurtosis, for example.
If you want to make a statement about the MEANS of group A vs group B, you need a tool that looks at means, not a tool that looks at the top percentile and nothing else!
This is the point Larry Summers made all those years ago. It did him little good, and I expect it will make little impression on most readers even today.
Name, I'm sorry I spun you up with my little snark. I seldom use emoticons to shout SNARK; I hope my readers will recognize the irony when I post it.
My point was the same as yours, in fewer words. I'm sure you recognize that a great many folks who read the information Kevin posted will immediately jump to the conclusions in my first paragraph, without bothering to follow my suggestion in my second paragraph. It's a fact of life and sadly, all the explanations from people like you and me will be as useful spitting in the ocean.
irony is dead
The point is not snark or irony, it is ACCURACY.
I don't give a fsck about the political points people want to score either pro or against white/asian/men/women. I just ask that, as STEP ONE, they actually understand WTF they claim to be arguing for or against.
Those who tore down Larry Summers couldn't get this correct. I don't expect most of those on the right to do any better.
he was accurate. But I'll give your words the consideration they deserve.
Achtually, no one here is being accurate...
You're all writing about "inventors" when what is charted is "people who have patents".
Which is not invention, it is a legally protected concept of granting privilege over an idea. Maybe the chart is just telling you who has the easiest route to getting a patent, or greatest incentive to obtains one and it's entirely a legal or cultural thing being measured.
There is no real mystery here. USPTO and the courts decided that software was patentable, and there are a lot of Asians employed in the software business. Many, if not most, of the patents are crap that nobody ever expects to assert so they never have to stand up to a challenge.
So what do you think the numbers would look like if we just looked at non-software patents?
I don't know but I found on "ipwatchdog" that 63% of the US patents granted are software or "software related". I suspect that Asians would still be over represented, companies go through the visa hassles for a reason, but I don't think it would be nearly so dramatic.
Actually, _you're_ the one who's inaccurate, most probably because the point flew way over your head. Just about everyone here is statistically literate to some degree and by that I mean they know perfectly well how to generate such curves. Do you take my point or did it fly right over your head?
Which “top 1%”?
I am a scientist with plenty of patents, and know many more people like myself. Unless we start our own business and it succeeds, we do not get rich. Even if we climb the corporate ladder and end up managing a lab and a few dozen people, we earn less than your dentist. Your more typical scientist earns between $100k and $150k.
On the theory the more inventions the better we need more immigrants in our colleges and more immigrants staying on after graduation. Incidentally they'll also fill the empty slots at many colleges.
I wonder what this would like if you stripped out software inventions.
Highest foreign born and non-white percentages, and one of the lowest fields for women--looked at "information".
What I didn't see in the report, though could have overlooked it, was the number of patents in each field over the course of the study.
What is to be done?
Does Kevin have any thoughts?
Why does something "need to be done"?
You really didn't get the point either.
Relative to whites (while ignoring Asians) aren't Blacks, Hispanics, and Native Americans actually doing well? Considering the sizes of their populations compared to Whites, it seems they are doing better than Whites. I may be missing something.
Strange though, having worked as an engineer for decades -- Blacks, Hispanic, and Native Americans have disproportionately small representation in the engineering work force. So I don't see how KD's chart is right.
Also consider that Native Americans are just 1% of the US population, but they get 17% of all patents? Again, seems not right.
You did indeed miss something. The figures already adjust for population share. Here's a more explicit quote from the paper:
Figure 5 summarizes these changes by showing, for each race and ethnicity group, the ratio of the share of the inventor population in 2000 and 2016 to the group’s share of the 2010 population. In both years the Black, American Indian or Alaska Native Alone (AIAN), Hispanics, and individuals with two or more race categories are under represented relative to their population share.
I think most patents are granted to industrial labs and other operations, that is they are typically team efforts. A company doing any kind of research or invention will patent anything that looks potentially profitable. The patents will be in the name of the workers, but they sign over rights to the company. So this may have more to do with hiring policies than individual talent or effort.
Correct. African-Americans, Latinos and women are underrepresented in scientific, engineering and technical employee populations, who do most of the inventing. Obtaining a patent takes significant investment, so for independent inventors, wealth is a factor - and the same groups are less wealthy on average than white males.
I am a white male engineering psychologist* with a handful of patents, all paid for by a corporate employer.
*PhD in experimental psychology, career spent in R&D of human operation and maintenance of complex systems.
What of patents held by corporations? Were they able to determine who did the actual inventing?
Most issued patents that have economic value are owned by companies. Inventors for these patents are almost always employees of those companies. Their employers generally have clauses in employment agreements that anything an employee invents is automatically the property of the company. The companies generally keep detailed track of who's invented what, because the law requires accurate disclosure of the individual inventors, who are required to be listed on the patent application, even if their employer owns any issued patent from the application.
Correct in all particulars.
Personally, I found this part disturbing:
Amazingly they link data on inventors from patents to anonymized, person-level identifiers, known as Protected Identification Keys (PIKs) which includes things like earnings and employment and they link that data to data on firms.
Privacy's corpse has been 3-d modeled for high-speed automated production and defilement.
Though it seems they are closing ground well.
Got a time-series for the other demographics?
For many black people, I think they struggle with the feeling that it doesn't matter what they do because they are going to get screwed in the end. Heard this story on NPR awhile back, and yes it was from the 1920s, but it was revealing.
https://news.harvard.edu/gazette/story/2020/06/the-1921-tulsa-race-massacre-and-its-enduring-financial-fallout/
You think blacks have the same hardships now as they did 100 years ago?
I know that I still express the cultural ideals that my parents instilled in me, and that they inherited from parents who grew up in the great depression. I also know that black people still deal with slights and assumptions pretty much daily. That will work your mind.
'The past is never dead. It's not even past.'
Response to an innocent bystander who just happened to be caught in the middle of turf war when the shooting started:
'Geeze, you were lucky to be shot in the arm instead of the brisket.'
'No, being shot int the arm is not a good thinng. And being where I was is not good luck.'
Unthinking callousness is a common trait in this particular group.
This is the original article
https://link.springer.com/article/10.1007/s10887-014-9102-z
Perhaps it's time for these types of report to parse Asian out a bit. Especially with two of the largest populations out there entering US workforce. Consider there are about as many immigrants from India as there are Native Americans.
From what I saw within the IT industry cranking out patents has been standard and the opportunities great, as the fields expanded rapidly. The same companies that leverage imported labor which often transition to more permanent residency. Primarily from India.
Seems possible selective immigration may be an influence on the numbers associated with Asian.
Kevin should know better.
isn't it the result of decades of H1B hiring among the Tech and Health care industries? Instead of investing in US education why not poach already educated people in countries with fewer opportunities at the time.?
This trend had a very negative impact on the emergence of a scientific native population in the US. Instead of having rising salaries that would attract native students, companies have imported huge numbers of foreigners for decades.
Not just H1B, immigration skews toward the educated. The top 1% of China and India combined is 28 million people. Give us your over-achievers and over-caffeinated!