Ulysses S. Grant is a classic late bloomer. Maybe you could argue the ability was always there, but it needed the right circumstances to become apparent. I think there is also an argument that he was ASD, and this may have contributed to his lengthy period of being overlooked, and difficulties in advancing himself in the military and business worlds.
Fascinating! Now I'm going to explore your archive. There must be some interesting parallels with entrepreneurs. For every Zuckerberg there will be someone who needed many years of experience to build a truly great business. Intellectual capital growing with the physical balance sheet.
Essential to late blooming are languages, as the Chech say "through learning a new language one gets a new soul". Reading, reading, reading and remembering most of it increases the mix of knowledge.
But to read a lot you need time and time comes with getting older.
HV Morton became famous in the second half of his life as a great travel writer. I've got all his books on a special shelf, and the beauty of his writing about England between the Wars will break your heart from a sense of a lost civilisation. But he was a mediocre journalist - famous mostly for being on the spot when Carter opened the tomb of Tutankhamun - before having a kind of conversion experience. He fell seriously ill while on assignment in Jerusalem and he "made a deal" with God that if he recovered and was able to go home to England, he'd travel all over the country and write about everything. He recovered from his illness and did exactly that, and his life took a complete turn ever thereafter. https://en.wikipedia.org/wiki/H._V._Morton
I will also add that Thomas Sowell as a late-bloomer. He has published close to 50 books in his lifetime (most recently in 2020), but he started getting his books published at the age of 40. He was a high-school dropout - but after the military service he completed his degrees at Harvard, Columbia, and Chicago.
"A study of when scientists publish their highest impact work found that scientists are just as likely to do their best work old as young."
That's what the study claims, but I'm not sure their model supports it. It does not include any field-specific estimates or adjustments (or at least, the authors refused to discuss it when I brought it up on Twitter with them and claimed that modeling individuals would account for this, even though I can't see how that possibly could: https://twitter.com/gwern/status/1170135752103137280 ), and so it doesn't refute Jones or the others. If you lump together life trajectories from researchers in fields like quantum mechanics where people are winning their Nobels in their 20s with trajectories from researchers in fields like medicine where they may be in their 60s, what will you get? Seems like you'd get something like no age-related effects...
Thanks for this. It is entirely possible I am getting some of this wrong. I am not qualified to critique the model they used. (I perhaps should have made that more clear in the piece itself.) But I would like to understand it better. I have some queries based on your comment...
Is the Jones paper you mention this one (https://www.pnas.org/content/108/47/18910, 'Age dynamics in scientific creativity', Jones andWeinberg) showing that cross-field differences in age are less substantial than cross-time differences, and that the age of achievement has shifted over time, but that it still varies somewhat by discipline?
I guess you would therefore argue that someone like Alexander Fleming, for example, isn't actually a 'late bloomer' because the average age when the work of a medicine Nobel gets done is 50+ anyhow? Whereas if he had been a Physicist in the early C20th then he would be 'late'.
This might be a dumb way to ask this question, but here goes... Can the results from Jones that the average age of achievement varies within disciplines for Nobel winners, and the results from Sinatra et al be right? Sinatra et al are looking more broadly at impact/citations not Nobel winners. Jones looks at breakthrough work but they look
broader than that. So maybe you can't be a Physics Nobel aged 50 but you can still have impact?
If that is true, could it be that there are subfield average age differences for Nobels (per Jones), but the probability Sinatra et al report of someone being 'successful' at any age still holds over time? Or is that not possible? Sinatra says 'the dependence of Q on exogenous factors, such as... dynamics of subfields... re- mains unknown.' That could be consistent with Jones, no?
It seems possible that there are explanatory factors about people who produce later even if Sinatra is wrong about age of scientific achievement.
For example, outside of the sciences it might be more difficult to measure the point of success, such as for politicians and writers. There's not quite the same thing as a paper with citations. But the general theory that you need some combination of Q, persistence, and luck, independent of each other, to be a successful politician appears to have some validity when applied to examples.
That was what I thought, even though they spun it otherwise. Assuming I understood the model they used - I'm not sure I understand their work. But then, I'm not sure they understand it either: I quoted that bit about "dynamics of subfield" to Sinatra as well since it seemed to contradict their other statements claiming that Jones-like differences are handled (in addition to no subfield variables seeming to appear in the model specification), and they ignored it and said to email them (which in my experience with academics means roughly https://www.reddit.com/r/ProgrammerHumor/comments/8agwx1/am_i_doing_this_right/ ) so I dropped it there. Presumably at some point someone will redo it.
> I guess you would therefore argue that someone like Alexander Fleming, for example, isn't actually a 'late bloomer' because the average age when the work of a medicine Nobel gets done is 50+ anyhow? Whereas if he had been a Physicist in the early C20th then he would be 'late'.
Yes, something like that. If it is normal for the best work in a field to be done late in life, perhaps because it requires gradually assembling a vast armada of empirical experiences & facts or traveling to a very distant research frontier, the concept of a 'late bloomer' doesn't really seem appropriate, as it's surely about designating some sort of outlier from an earlier norm? You wouldn't call sequoia cedars or bristlecone pines 'late bloomers', would you?
> So maybe you can't be a Physics Nobel aged 50 but you can still have impact?
I would bring in Simonton's equal-odds rule here. The big picture *seems* to be something like, each piece of research output has roughly equal odds of being impactful, however you measure that, like Nobel-Prize-winning (perhaps because we're really bad at predicting impact and Nature keeps surprising us about where any given topic can lead to?), but that lifecycle effects come from quantitative output differences. In the same way that Euler and Gauss produced a *lot* of work and were astoundingly prolific, and this tends to be true of all the major figures, peak ages of impact are peak ages of total output. Field specific age effects might relate to restraints on output. We could tell a story about young mathematicians/physicists being so prolific because raw fluid intelligence leads to manipulating new abstractions in all sorts of fertile ways starting from a blank sheet of paper, while a doctor has no such luck as his materials are much more uncooperative and messy.
On Simonton, and pine cones, perhaps my general point stands that 'late' bloomers are often in fact people working with concrete intelligence and lifelong learning... that obviously shows up in fields like medicine etc. So when we see other so called late bloomers we might think more about recategorising the type of artist/writer etc that they are rather than just as 'late' e. g. Frost did some excellent late work as well as young work so we might want to view him as doing something more akin to medicine, while many other poets peak like physicists. The theoretic/experimental distinction seems to be somewhat consistent across science and arts.
At some point this just reverts to 'there's huge variation and individual difference is the biggest factor'. Maybe I ought to leave it there.
Brilliant insight.. Such affirmation so delicious to my soul.
Ulysses S. Grant is a classic late bloomer. Maybe you could argue the ability was always there, but it needed the right circumstances to become apparent. I think there is also an argument that he was ASD, and this may have contributed to his lengthy period of being overlooked, and difficulties in advancing himself in the military and business worlds.
Ah great, thanks so much!
Fascinating! Now I'm going to explore your archive. There must be some interesting parallels with entrepreneurs. For every Zuckerberg there will be someone who needed many years of experience to build a truly great business. Intellectual capital growing with the physical balance sheet.
Absolutely. I think the average age of a tech entrepreneur is 45. Seems like an under explored area.
Essential to late blooming are languages, as the Chech say "through learning a new language one gets a new soul". Reading, reading, reading and remembering most of it increases the mix of knowledge.
But to read a lot you need time and time comes with getting older.
HV Morton became famous in the second half of his life as a great travel writer. I've got all his books on a special shelf, and the beauty of his writing about England between the Wars will break your heart from a sense of a lost civilisation. But he was a mediocre journalist - famous mostly for being on the spot when Carter opened the tomb of Tutankhamun - before having a kind of conversion experience. He fell seriously ill while on assignment in Jerusalem and he "made a deal" with God that if he recovered and was able to go home to England, he'd travel all over the country and write about everything. He recovered from his illness and did exactly that, and his life took a complete turn ever thereafter. https://en.wikipedia.org/wiki/H._V._Morton
Oh, I checked and discovered he was only 34 when he wrote In Search of England, which seems absurdly young to me now.
Awesome post!!
Thanks! You’ll enjoy the book
I will also add that Thomas Sowell as a late-bloomer. He has published close to 50 books in his lifetime (most recently in 2020), but he started getting his books published at the age of 40. He was a high-school dropout - but after the military service he completed his degrees at Harvard, Columbia, and Chicago.
The element or factor Q is too vague to explain anything. Persistence and luck have more explanatory power.
"A study of when scientists publish their highest impact work found that scientists are just as likely to do their best work old as young."
That's what the study claims, but I'm not sure their model supports it. It does not include any field-specific estimates or adjustments (or at least, the authors refused to discuss it when I brought it up on Twitter with them and claimed that modeling individuals would account for this, even though I can't see how that possibly could: https://twitter.com/gwern/status/1170135752103137280 ), and so it doesn't refute Jones or the others. If you lump together life trajectories from researchers in fields like quantum mechanics where people are winning their Nobels in their 20s with trajectories from researchers in fields like medicine where they may be in their 60s, what will you get? Seems like you'd get something like no age-related effects...
Thanks for this. It is entirely possible I am getting some of this wrong. I am not qualified to critique the model they used. (I perhaps should have made that more clear in the piece itself.) But I would like to understand it better. I have some queries based on your comment...
Is the Jones paper you mention this one (https://www.pnas.org/content/108/47/18910, 'Age dynamics in scientific creativity', Jones andWeinberg) showing that cross-field differences in age are less substantial than cross-time differences, and that the age of achievement has shifted over time, but that it still varies somewhat by discipline?
I guess you would therefore argue that someone like Alexander Fleming, for example, isn't actually a 'late bloomer' because the average age when the work of a medicine Nobel gets done is 50+ anyhow? Whereas if he had been a Physicist in the early C20th then he would be 'late'.
This might be a dumb way to ask this question, but here goes... Can the results from Jones that the average age of achievement varies within disciplines for Nobel winners, and the results from Sinatra et al be right? Sinatra et al are looking more broadly at impact/citations not Nobel winners. Jones looks at breakthrough work but they look
broader than that. So maybe you can't be a Physics Nobel aged 50 but you can still have impact?
If that is true, could it be that there are subfield average age differences for Nobels (per Jones), but the probability Sinatra et al report of someone being 'successful' at any age still holds over time? Or is that not possible? Sinatra says 'the dependence of Q on exogenous factors, such as... dynamics of subfields... re- mains unknown.' That could be consistent with Jones, no?
It seems possible that there are explanatory factors about people who produce later even if Sinatra is wrong about age of scientific achievement.
For example, outside of the sciences it might be more difficult to measure the point of success, such as for politicians and writers. There's not quite the same thing as a paper with citations. But the general theory that you need some combination of Q, persistence, and luck, independent of each other, to be a successful politician appears to have some validity when applied to examples.
I hope this is clear...
> That could be consistent with Jones, no?
That was what I thought, even though they spun it otherwise. Assuming I understood the model they used - I'm not sure I understand their work. But then, I'm not sure they understand it either: I quoted that bit about "dynamics of subfield" to Sinatra as well since it seemed to contradict their other statements claiming that Jones-like differences are handled (in addition to no subfield variables seeming to appear in the model specification), and they ignored it and said to email them (which in my experience with academics means roughly https://www.reddit.com/r/ProgrammerHumor/comments/8agwx1/am_i_doing_this_right/ ) so I dropped it there. Presumably at some point someone will redo it.
> I guess you would therefore argue that someone like Alexander Fleming, for example, isn't actually a 'late bloomer' because the average age when the work of a medicine Nobel gets done is 50+ anyhow? Whereas if he had been a Physicist in the early C20th then he would be 'late'.
Yes, something like that. If it is normal for the best work in a field to be done late in life, perhaps because it requires gradually assembling a vast armada of empirical experiences & facts or traveling to a very distant research frontier, the concept of a 'late bloomer' doesn't really seem appropriate, as it's surely about designating some sort of outlier from an earlier norm? You wouldn't call sequoia cedars or bristlecone pines 'late bloomers', would you?
> So maybe you can't be a Physics Nobel aged 50 but you can still have impact?
I would bring in Simonton's equal-odds rule here. The big picture *seems* to be something like, each piece of research output has roughly equal odds of being impactful, however you measure that, like Nobel-Prize-winning (perhaps because we're really bad at predicting impact and Nature keeps surprising us about where any given topic can lead to?), but that lifecycle effects come from quantitative output differences. In the same way that Euler and Gauss produced a *lot* of work and were astoundingly prolific, and this tends to be true of all the major figures, peak ages of impact are peak ages of total output. Field specific age effects might relate to restraints on output. We could tell a story about young mathematicians/physicists being so prolific because raw fluid intelligence leads to manipulating new abstractions in all sorts of fertile ways starting from a blank sheet of paper, while a doctor has no such luck as his materials are much more uncooperative and messy.
This is so helpful, thank you.
On Simonton, and pine cones, perhaps my general point stands that 'late' bloomers are often in fact people working with concrete intelligence and lifelong learning... that obviously shows up in fields like medicine etc. So when we see other so called late bloomers we might think more about recategorising the type of artist/writer etc that they are rather than just as 'late' e. g. Frost did some excellent late work as well as young work so we might want to view him as doing something more akin to medicine, while many other poets peak like physicists. The theoretic/experimental distinction seems to be somewhat consistent across science and arts.
At some point this just reverts to 'there's huge variation and individual difference is the biggest factor'. Maybe I ought to leave it there.