←back to thread

1401 points alankay | 1 comments | | HN request time: 0s | source

This request originated via recent discussions on HN, and the forming of HARC! at YC Research. I'll be around for most of the day today (though the early evening).
Show context
david927 ◴[] No.11940304[source]
You have stated before that the computer revolution hasn't happened yet. It seems we stopped trying in earnest back in the early 1980's. Why?

And what could be done to re-spark interest in moving forward?

My gut feeling says that it would require a complete overhaul in almost every layer in the stack we use today and that there's reluctance to do that. Would you agree to some degree with that?

replies(1): >>11946023 #
alankay1 ◴[] No.11946023[source]
Have you seen the "gyre" in the Pacific?
replies(1): >>11946673 #
david927 ◴[] No.11946673[source]
Yes, of course; we agree. Let me refine my question:

In our not-quite-an-industry, we seem to laud attempts to optimize the artifact of a residual hack, and we are absolutely dismissive of attempts to rebuild the stack as being too ambitious. And the problem with being dismissive is that it's a judgement without trial. We have precious few "crazy professors" and no tolerance for them.

What can we do?

There's the 2020 group in San Francisco. Is that kind of meetup the right direction?

replies(1): >>11947083 #
alankay1 ◴[] No.11947083{3}[source]
"Cops need criminals"

"Doctors need disease"

Things have happened in the past when passionate and confident people with chops have decided to do something. These have not always correlated with "What is actually needed" (often not) but they usually get things into a state where people who are mainly trying to advance their own goals see some advantage (including "tech chic").

We are certainly not in a position where this can't happen a few more times.

Looking back, I've been struck not by how few really good researchers there are, but more so by how few really good managers of researchers there have been, and even more so by how really really few good funders there have been.

Maybe too simplistic, but in my view great funding has caused great stuff. So the funders should get the gold medals rather than the researchers! (Think about it: the good funders give out the gold in advance knowing full well that if even if they are very luckly 70% of the gold will turn to lead in just a few years!)

replies(1): >>11950533 #
david927 ◴[] No.11950533{4}[source]
few really good managers of researchers

Bob Taylor has been generally praised as a great manager, and I believe them of course. But when I heard stories of his management style, it seemed to go against every instinct we have on how to foster creativity. Could you comment on that?

Also, in terms of funding, I wonder -- haven't things changed? Wasn't funding more important in, say, the 1960's and 70's due to the cost of computer time, especially at the processing level that would let you "see the future." A $1000 computer today is not so different, in terms of power, from a similar machine 5 years ago, right?

Things have happened in the past when passionate and confident people with chops have decided to do something.

But wasn't the past more open? When a field is just forming, everything is crazy, so nothing is. It's only when it has solidified (and in exactly the wrong direction) that you would more expect to find a "crazy professor" having more of an impact, right?

replies(2): >>11951281 #>>11953053 #
mmiller ◴[] No.11951281{5}[source]
As Alan has said, the importance of funding is more about finding the problem. There's a lot of wandering and false leads involved (though, finding out the false leads is nevertheless valuable. It's still new knowledge). In terms of hardware, the thing to anticipate is what computing power will be necessary for accomplishing things decades from now, and paying the bill for that, not considering what one can buy from a computer retailer today, because that doesn't create the excuse to think of what computers will be able to do.
replies(1): >>11951868 #
1. david927 ◴[] No.11951868{6}[source]
Using funding to find the problem is inefficient; side projects are a much more distributed and efficient method and they don't need funding.

As for hardware, you're repeating what's clear. What I'm saying is that you used to have to purchase the future but now you can mimic those coming changes and, while slower, offer perfectly acceptable performance.

I guess I'm trying to lead Alan to say something to the effect that the next PARC can be groups, working together, on their own time, over Skype. I'm not sure he believes that but I do.