Dear Marc Andreessen

Woah, what? Are you really saying that the negative uses of technology in the World Wars have outweighed the positive uses of technology (even if we limit ourselves to a timeframe that contains an unusual amount of war deaths)? It's beyond ridiculous to reduce the benefits of technology to "some guy" getting an artificial heart, the moonwalk (which depending on your perspective is either a prestige achievement or a step towards space exploration whose benefits we haven't seen yet), and a messaging app. I really, really hope you're being disingenuous instead of just ignorant because I'm practically rendered speechless at the thought of having to explain how electrification alone has saved countless lives.

OTOH, I guess I shouldn't be that surprised, since it seems to be a very common phenomenon for people to ignore steady long-term effects in favor of flashier one-offs, even if the latter has a tiny, tiny, tiny cumulative effect in comparison.

As just a couple of examples (I tried to focus on things that would be the most affected by technological/scientific/economic progress):

- The five-year mortality rate of breast cancer today is roughly the same as the mortality rate of GIVING BIRTH in 1900 (also known as something damn near half the population does, several times in their life).

- The Spanish flu infected 500 million and killed 100 million people in 1918. Today you can go to Safeway and get a flu shot for 10 bucks.

- In the early 20th century, the life expectancy at birth was 31 years of age. In 2010 it was 67 years.

- Per CDC data, 60 years ago, 38000 people died of polio each year in America alone. These days, 300 people die of polio per year in the ENTIRE WORLD.

- From 1920 to 1980, 395/10000 people died from famine each year. In 200, that number was THREE out of every 10000.

Of course I agree with the larger point that's tangential to the one you're making: blind progress without ethical safeguards is definitely foolish. That's not really what this discussion is about at all though (there are people already talking about the need for ethical standards around AI development and I'm the first to agree with that).

> America was better in the fifties, middle class wise, when unions were strong, compared to what it's now.

This is so incredibly ignorant of history that it's unbelievable. You're aware there was a World War right before the 50's, right? And damn near the entire world was either coming out of centuries of suppressed economies under colonialism or trying to recover from being, you know, blown up? Might that have _anything_ to do with the success of American labor (it's not a coincidence that the opening of developing-world markets in the 70s and 80s coincided with the loss of America's unskilled-labor competitiveness)? Not to mention the fact that "middle-class wise" is an incredibly idiotic metric to use as a proxy for "better". Refer back to the list I had above for examples of _real_ ways that lives can improve (or seriously, just Google it: you can literally find HUNDREDS of ways life was worse back then, even if you limit yourself to the 50s and to America). The places that we _have_ gotten much worse (i.e. economic prospects for lower and lower-middle classes) is amply addressed by the grandparent comment's allusion to a proper welfare state and basic income. To put it another way, retarding progress so people can have pretend-productive jobs could not be stupider; through inefficiency, you're destroying wealth that can be redistributed to people who actually need it instead of to people who happen to hold arbitrary obsolete jobs. What you're fundamentally saying is that we should implement welfare in the most inefficient way possible and then give it not to the poor, or the sick, or the needy; but to people in arbitrary industries (like dockworkers or taxi drivers) at arbitrary income levels.

Thanks for keeping on making my point :) There is no "we". There is you and some people that are in ideological agreement with you, but there is no we. Why do you have such a hard time understanding this?

Those past predictions tell us that we weren't very good at predicting the coming of strong AI in the past, but it doesn't necessarily mean that current predictions are unreliable.