Not my area of expertise, but I think the intuition for why evolution didn't make us all Einstein's comes in two parts:
(1) IQ is very polygenetic - a big IQ gene might be 0.1 IQ, and there aren't a lot even that large, if I recall the large genome wide association studies done on this. As GWAS studies get larger we find more, but they're all super small effects.
(2) Having an extra 0.1 IQ point is easy to lose in the noise of other effects on having offspring.
So even if we assumed IQ was perfectly correlated with more surviving offspring in modern societies, it might take an extremely long time (even longer than a whole planet and a thousand years) to make a detectable difference.
I've been an engineer at Periscope for the past couple years and would be happy to answer any questions about the company or life as an engineer here.
I've worked on the whole stack here, but my favorite parts have been new charting features (most recently annotations), client and server side latency (I love the chrome flame graph profiler), machine learning for lead scoring models (hooray for scikit), and anytime we can ship a customer feature request the same day they request it.
Send me an email at [email protected] if you'd like to talk about life at Periscope! :)
The system isn't training on antonyms and analogies - it's training on wikipedia. It's learning the meaning (and multiple senses) for every word it can find.
The test they use to see if it actually learned what these words meant, in a limited sense, is to test it against a subset of verbal IQ tests (not what it was trained on!). You could ask it the antonym, synonym, or analogy for anything in English. This is an extension of word2vec / word embeddings.
That it beats the scores of college graduates impresses me.
"it's training on wikipedia. It's learning the meaning (and multiple senses) for every word it can find."
I don't think that is entirely correct. After cursory reading of the paper, my understanding is that they look up a list of word senses for each word in a dictionary (or multiple dictionaries). And then they try to learn something about each of those word senses from wikipedia (that is they create seperate word embeddings for each of those senses). So what they do not do is to learn what senses a word has. That is done by the humans who created the dictionaries.
What that means is that they cannot pick up new senses of words, which doesn't matter for answering IQ test questions because these questions rarely change and are typically based on well established word meanings.
Unfortunately it makes this approach less than ideal for things like understanding the news (something I'm working on), where new contexts of words keep popping up all the time.
Startups involve having to do everything with a tiny group of people, so the more skills you have personal experience with the better. If you business depends upon your product, sales and marketing skills, and you've only been a developer, it's going to be a very tough time.
Working at a startup is a great way to get some experience in several other domains, especially if you are one of the first employees. Even if you don't do it first hand, you'll get to spend a ton of time with "head of sales", "head of marketing", "head of product", because they'll be the other three dudes in your office.
One problem with time boxing is that it isn't the actual terminal goal - when trying to be self-motivate, there would be an extra level of abstraction between effort and reward, as you'd now be self-motivating to timebox, rather than to do anything inherently useful.
Besides, it fails the "too good to be true" heuristic. If it actually persistently improved productivity in a meaningful manner, you'd expect it to have been better adopted widely by now.
Something kind of close is www.beeminder.com where you can pick a goal, keep track of your progress, and have an auto-forfeit sum of money. It doesn't have verification built in, but it's part way there.
I have a poor sense of how I spend all my time, and that leads to sub-optimal decision making. I've been building a web app to easily track my time usage and display it as a historical calendar, and adding features to get the most insight out of reflecting on it (running stats, notes, etc.) The most recent handy feature was an xmpp chat bot interface for lowering the friction on adding data.
The inspiration was the quote "Those who cannot remember the past are condemned to repeat it", which jives with the observation that how one spent their time last month is a better predictor of their future than how they wish they would spend this month. People have a lot of inertia, but tend not to notice.
I'd be especially interested in any tips about research on past-time perception.
The only plausible interpretation I see is that the event had costs that were counted as liabilities, but that the revenue from the event was not yet recognized as assets.
(1) IQ is very polygenetic - a big IQ gene might be 0.1 IQ, and there aren't a lot even that large, if I recall the large genome wide association studies done on this. As GWAS studies get larger we find more, but they're all super small effects.
(2) Having an extra 0.1 IQ point is easy to lose in the noise of other effects on having offspring.
So even if we assumed IQ was perfectly correlated with more surviving offspring in modern societies, it might take an extremely long time (even longer than a whole planet and a thousand years) to make a detectable difference.