Live data from Hacker News

Ask HN: What should we fund at YC Research?

news.ycombinator.com

401–410 of 545 posts

Re: Ask HN: What should we fund at YC Research?

#401
post #396
post #365

Earlier quoted context omitted.

I’d have said AI safety, but it’s already on your list. Excellent. But, why? Nothing about "AI safety" is remotely useful. It would be like researching "alcohol safety" (something everybody has access to and nobody can restrict) or researching how ants can influence "human safety" (tiny things trying to influence massive things). It's nonsense no matter which way you think about it unless you just want to write SF no…

> boost corporate profits while paying highly skilled professional volunteers nothing. This is a problem of ancient outdated patent and copyright law, not scientific innovation. Its in the field of politics since the industry is entirely built on state enforced IP legislation.

But open source code is all about "no patent restrictions" and free licensing even with retained copyright (or even outright copyright assignment/dumb CLAs).

Iterating those concepts don't seem to fix the issue of Twitter running 50,000 copies of an application developed by 3 open source volunteer people and not paying those 3 people anything.

Re: Ask HN: What should we fund at YC Research?

#402

Earlier quoted context omitted.

> Humans need more lifespan, badly. Do we? Or do we need more realistic perspective about death? Why quantity over quality? Seems there's a lot of quality improvements that could be made [1]. For instance, I read recently (can't recall source) that San Fransisco's homeless is the least happy in the world. Also that the top 1% in the world make over $35K a year, but there are far happier people than exist in middle Am…

I remember when I first deconverted. I was horrified that there was no heaven, and that my loved ones weren't waiting for me. That they were just gone forever. They didn't exist. They weren't happy, and I would never see them again. I see many atheists carry on religious memes after they deconvert. Which is only natural, religion isn't just a single idea, it's deeply ingrained into our culture. One of those memes is…

You've joined the company of great people in history like Tsiolkovsky.

Re: Ask HN: What should we fund at YC Research?

#403
post #334

Earlier quoted context omitted.

> Or do we need more realistic perspective about death? We do: we need to realize how horrible it is that we treat it so casually, rather than as a tragedy. Perhaps then we'll do something about it. > [1] Notice that cancer, heart-disease, Alzheimer, etc research also falls into this category - allowing people to live with less suffering. All three of which (especially the first and last) have heavy ties to aging. It…

Wanting to live forever has to be the apex of egoism and arrogance and all that is wrong with the human mind. Death is good because it leaves room for new people with new ideas. Being against death reveals a profound misunderstanding about what life is about, which is: finiteness, urgency, novelty, creativity.

Not everyone has to share your profound insight about the meaning and purpose of life.

This is a classic example of a limited rationalised mindset. Imagine how much more enriched people lives would be if they got to know their great grandparents and more. Death is a tragedy, there is nothing stopping you and others who share your beliefs from ending life at any moment they want. This is about options, think about the potential -- it would change society is so many ways.

Re: Ask HN: What should we fund at YC Research?

#404
Here's the other medical condition in dire need for intelligent solutions. See my other post for details. https://news.ycombinator.com/item?id=11002063

Acoustic Neuroma:

This is a benign tumor that grows around the auditory nerve between the brain and ear organs. It squeezes the nerve to the point where hearing on that side is eventually lost. This leads to SSD (single-side-deafness) which can be a debilitating condition.

One does not realize just how useful having two working ears happens to be. SSD causes two significant problems.

The first is the loss of the ability to locate the source of sounds. Your kid calls for help? You have no clue where they are. Someone calls you at the office? Same thing. You literally can't locate sounds.

The other problem is the inability of the brain to run its "noise reduction" routines on sounds coming in. I know people with SSD who, in a busy restaurant, literally can't hear the person sitting in front of them. The brain is able to process sound from both ears and allow the listener to cognitively focus on a desired source. It's a mini phased array with a powerful processor. Take away one of the two available microphones and the loss of capabilities is significant.

Another potential side effect of Acoustic Neuromas is that they can affect the facial nerve for one half of your face. If this happens, the muscles on half the face go limp and the entire half of your face droops.

Current "solutions" don't guarantee the preservation of hearing, hearing recovery or the preservation/protection of the facial nerve.

The first "solution" is to drill a ~ 1 inch hole on the skull right behind the ear and remove the tumor. Of course, the nerve is cut out with the tumor. So, SSD forever. The hole is filled with fat. A metal plate is used to cover the hole. A stud is screwed into your skull and a little box handed to you.

This box is a bone conduction transducer. It has a microphone and transducer that will vibrate your skull. The idea is that your skull will conduct this sound from one side to the other so you can hear sounds from both sides. Of course, you lost the phased array, so hearing this sound, while perhaps useful, might just add to the noise. And, of course, it probably won't be full fidelity. You get to walk around with a pager bolted to the back of your skull [1]. There are other, less invasive, solutions [2] but none that fix the underlying problems. The BAHA devices mean you have this stud --which looks like the end of a spark plug-- screwed into your skull. Skin infections and other problems are some of the issues patients have to deal with.

Believe it or not there's a whole building in Los Angeles dedicated to the above-described butchery. I believe it's run by a USC professor (or ex, don't remember). I believe the procedures start somewhere around $50K.

The next "solution" is endoscopic surgery. Not as brutal. They drill a 1 cm hole on your skull, go in and chip away at the tumor. Practitioners can be more careful and do their best to avoid damaging the acoustic and facial nerves. Yet, nobody will guarantee this. Complete loss of hearing is very likely and facial drooping could happen just as well. While less intrusive than a huge hole and a metal plate, this surgery does not seem to offer improvements in results. You are still likely to be offered a BAHA bone conduction device as an add-on. My guess is this procedure runs somewhere in the $80K+ range.

The third approach is to use Gamma Ray Surgery. This is commonly used for Cancer patients. The tumor is mapped via MRI. The patient's head is then bolted to a table under the Gamma Ray machine and the tumor's blood supply is zapped. The goal is to kill it and keep it from growing. Due to dose limits the treatment takes several weeks. I don't really know how much it costs.

This approach does not invade the cranial cavity, which is good --you can still go scuba diving. It can be precise enough that the nerves are not touched at all. In other words, some hearing could come back and facial feeling/control as well. Yet, nobody will guarantee any of it. Part of this is due to medical liability laws in the US, a whole other subject. BAHA devices might still be offered.

What is needed for people afflicted with this condition could take many forms:

- Early detection: This is a slow growing tumor. Can it be predicted through DNA or other testing? Is there an easy way to find markers and attack the tumor when it is at the 2 mm stage rather than waiting for the patient to notice hearing loss only to find a 3 cm tumor in their skull?

- Better treatment: Can Gamma Ray (or some other non-invasive approach) be improved/perfected in such a way that the tumor can be killed off more effectively while avoiding damage to the nerves and perhaps even restore natural hearing?

- Better devices: Surely there has to be a better and more intelligent approach to restoring some sound perception than bolting a stud into the patient's skull or shoving an uncomfortable lump of plastic into their ear.

[1] https://en.wikipedia.org/wiki/Bone-anchored_hearing_aid

[2] http://www.transear.com/

Re: Ask HN: What should we fund at YC Research?

#405
post #235

Globally open borders. Allowing people move freely (globally), following opportunity, seems like the best way to reduce global inequality. We allow everything else to move quite freely but if you've ever dated a foreigner or god forbid had to flee home you will understand the struggles people face when trying to move across restricted borders. It seems like a study into how globally open borders would affect inequali…

What are the positive effects that you expect? Saying "reducing global inequality", do you mean something like communism, where everybody shall be equal? Please, don't take my question negatively. I have myself lived in a communist country and I think it had both good and bad parts. Essentially it boils down to whether everybody becomes equally miserable or equally happy. Another problem that I could see with globall…

The goal is not to dissolve but for everyone to respect and accept. Canadian Prime Minister Justin Trudeau puts it perfectly in this video: https://www.facebook.com/quartznews/videos/1097902360243465/

Re: Ask HN: What should we fund at YC Research?

#406

Stack Risk. We need to know how much CPU power entities have at their dosposal. If statistics aren't available, find data centers and energy breakdowns backout those costs and use sqftage to find amount of servers. COmputing power growth rate is likely biggest metric tp assess how the tech industry is doing and possibly better at measuring an economy than GDP. The risk asdociated with having elephant flow like owners…

Sounds analogous to understanding an electricity grid and power generation. Distributing CPU power (decentralizing it) by having more people run servers in their homes is the Internet equivalent to rooftop solar.

Understanding how the electricity grid works is non-trivial, maybe I am mischaracterizing your suggestion that it is not.

Digital Ocean, a smaller provider, has 64gb memory 6 core intel cpu in a single server[1]. These are wired up into a cluster or clusters and probably plug in to a backbone provider like level3 or a big ISP. Small fish, last time i looked they have a half dozen or so regions with a few zones.

AWS tries to limit data centers to 100k servers, most being between 50-80k servers per data center.with a power capacity of 25-40megawatts[2].

So no, I don't think you could just get a few people to run servers in their house and offset that.

You also connect through an ISP. ISP can turn your connection off. Google is an ISP, runs popular DNS name servers, sells web donains, sells certificates, etc. so, they have a business at every level of the layer from hardware to application.

I guess I just take the positiom that we should know what percentage of power, routing and flow these guys control.

[1]https://www.digitalocean.com/community/questions/what-are-th...

[2]http://datacenterfrontier.com/inside-amazon-cloud-computing-...

Re: Ask HN: What should we fund at YC Research?

#407
post #334

Earlier quoted context omitted.

> Or do we need more realistic perspective about death? We do: we need to realize how horrible it is that we treat it so casually, rather than as a tragedy. Perhaps then we'll do something about it. > [1] Notice that cancer, heart-disease, Alzheimer, etc research also falls into this category - allowing people to live with less suffering. All three of which (especially the first and last) have heavy ties to aging. It…

Wanting to live forever has to be the apex of egoism and arrogance and all that is wrong with the human mind. Death is good because it leaves room for new people with new ideas. Being against death reveals a profound misunderstanding about what life is about, which is: finiteness, urgency, novelty, creativity.

I think defining other people to be "profound[ly] misunderstanding what life is about" just because they do not believe the same as you when it comes to something so fundamental and personal as one's life is much closer to "the apex of egoism and arrogance and all that is wrong with the human mind".

We don't need room for new people with new ideas, room isn't the issue: as far as we know so far, the Universe is virtually infinite. When we come around to defeating death (it's bound to eventually happen) we'll start terraforming, doing generation-ships to reach even further, and the rate of growth of human knowledge will reach an even more unimaginably fast pace.

Re: Ask HN: What should we fund at YC Research?

#408
post #188

Education. Everyone knows it's broken,not effective anymore and literally a waste of time and money. More so if you had the oppurtunity to experience what education means in third world countries.Spoiler alert : it's a joke that will make you cry. It needs to be fixed as quickly as possible and research needs to be done for alternatives to help people in the process of learning.Online education is a good start, but w…

Education isn't ideal now, but I don't think online education is a good start or even a viable model. Training more and better teachers is the way to go, especially in "third world countries". Not only is there big issues with motivation and the whole social/interactive/distant aspect to moocs; what do you do when you don't have internet, or your device breaks? Now your education depends on a machine that costs a cou…

I believe education will find a natural middle-ground.

In many instances the benefit of having an in-person instructor has much to do with the social interaction/interpersonal support that they can give to a struggling/minimally motivated student. That said, motivated learners don't always need that kind of personal interaction and would benefit from self-guided learning opportunities. This is obvious, but is hard to manage well.

In the end, I believe students will learn along a fluid spectrum from 'one-on-one' to 'fully distributed online learning' - both of which will heavily lean on better tools.

There are tens of thousands of lessons, activities, individual interventions and curriculum modifications created every day by instructors in this country - but all of that knowledge and all those artifacts are effectively lost (as almost nothing gets distributed) and the same creative labor is repeated over and over.

So - we should be looking for better ways of mediating in-person instruction to both improve that kind of instruction and as an engine for the creation of universally accessible learning materials. Students who have the ability/proximity to a physical schools get the benefits of personal interaction, while students who cannot get to school (or are underserved by that school) can get the benefits of time shifted lessons/activities/etc. from anywhere in the world.

Re: Ask HN: What should we fund at YC Research?

#409
Mind uploading.

It's definitely far off. But it seems to me this could be an amazing shortcut/hack to achieve a lot of other things here: once we have mind uploading, we've basically conquered death. We can also probably figure out, at that stage, how to run the minds faster or with extra memory or something, increasing the standard human intelligence.

This is also a way to achieve "superintelligence" via human-based intelligence, rather than through artificial intelligence.

That would be where I'd bet my money.

Re: Ask HN: What should we fund at YC Research?

#410
post #368

Stack Risk. We need to know how much CPU power entities have at their dosposal. If statistics aren't available, find data centers and energy breakdowns backout those costs and use sqftage to find amount of servers. COmputing power growth rate is likely biggest metric tp assess how the tech industry is doing and possibly better at measuring an economy than GDP. The risk asdociated with having elephant flow like owners…

Stack Risk. We need to know how much CPU power entities have at their dosposal. That's actually a great idea and almost impossible to make reliable. We could measure "stack risk" for registered, 100% above board companies with their own data centers and employees, but we can't really measure the risk of some guy in eastern europe with access to a 1 billion node, globally distributed and illegal obtained, Android bot…

Correct. However, we could see who has a legal android botnet, and see how easy it is to take over.

Not to mention, if someone did have access to even 1million phones they could topple a cell network by spam sms amd calling to tie it up.

I think someone should look at it.

Post reply on HN