Ask HN: What is now on the fringe?

34 points by ninjamayo ↗ HN
I was watching a video of Steve Jobs the other night talking about the experience of the Homebrew Computer Club and how these guys were working on the fringe rather than mainstream. Steve Jobs was looking at the fringe for new ideas and innovation so I was wondering what do we see at the fringe these days? What are those ideas that are not mainstream today but could become the "personal computer" level of idea in a decade or two.

31 comments

[ 3.4 ms ] story [ 68.6 ms ] thread
Vector representation of symbols, concepts, terms, image chunks, thoughts or words along with the unique ways to construct the feature attributes for these vectors and calculations for similarity between them. Essentially, new ways of mimicking human cognition without built-in rules. See also Computational Theory of the Mind (CTM)
Open source architecture, as in real estate.
Just curious, why do you think this? And what would it take?

I am personally working on open source tools for architectural design data communication, hence my curiosity.

Can you send a couple of links to get an idea?
(comment deleted)
Vector representation of symbols, concepts, terms, image chunks, thoughts or words along with the unique ways to construct the feature attributes for these vectors and calculations for similarity between them. Essentially, new ways of mimicking human cognition without built-in rules. See also Computational Theory of the Mind (CTM) https://en.wikipedia.org/wiki/Computational_theory_of_mind
THIS is a great idea. THIS is how brain-wave-mapping should be approached. THIS is actually how AI will become anything.
Interesting: yes. Fringe? Not so much. Vector representations of linguistic entities (tokens, types, concepts) have been the foundation of information retrieval for many years. There are new ways of finding better feature definitions but vectors for encoding natural language information isn't a new idea.
Affordable housing solutions. We are increasingly pushing people out into the street. If you can really solve this, there is a huge market for it.
It could already have been solved at least in theory.

A Tiny House on Wheels or THOW can be built for between 5000 and 50,000 dollars depending on specifications. I have even talked to two people (on HN and on Reddit) on the verge of homelessness who are building small transports for shelter (several hundred dollars) which they can haul by bicycle. These can be very effective in the colder months.

The sticking point is Land, or rather, Parking. Tiny Housers are very political active and slightly devious but for people in cities in particular this is tough.

There is a nice technical problem to solve for here if you could build a closed loop system for each input and output for a small house. I don't know that the technology exists for getting it to work properly even on a small scale at reasonable cost.

Because if one could solve that, then lots of problems go away.

I am well aware of the tiny house movement. They were put on wheels to get around the fact that it is illegal to build a house that small. It is a legal loophole. It also means they are relegated to being glorified trailers, which means they will never be on even footing in many important respect with a house with a foundation.

Furthermore, because it is "cutting edge," it isn't yet accessible to most poor people. A lot of tiny houses are insanely expensive per square foot, containing quite high quality materials. They are basically aimed at eccentrics, not nirmal people with a limited income. For $50k, you could build a house that is 600-1000 sq ft on a foundation without a lot f bells and whistles.

The average new home today is nearly 2500 sq ft -- more than twice the size of the average new home in the 1950s -- and contains fewer people than the average house in the 1950s.

We don't need eccentric high quality trailers. We need normal houses with actual foundations that aren't trying to compete with the Taj Mahal for grandeur.

This isn't remotely solved and the primary areas that need to be addressed have nothing at all to do with construction technique. They primarily have to do with policy reform and financing mechanisms.

> It also means they are relegated to being glorified trailers, which means they will never be on even footing in many important respect with a house with a foundation.

Well sure. It means you might have have trouble getting post delivered to your door or getting utility hookups in a city, or trash collection.

However those kinds of things are much more easily hacked than this:

"This isn't remotely solved and the primary areas that need to be addressed have nothing at all to do with construction technique. They primarily have to do with policy reform and financing mechanisms."

I completely agree! Meanwhile we have a problem... Get back to me when it costs 5-10 years of average wages to purchase the average house. Until then Tiny Houses are at the least a good interim solution to not becoming homeless or massively indebted or ripped into by landlords.

> Furthermore, because it is "cutting edge," it isn't yet accessible to most poor people. A lot of tiny houses are insanely expensive per square foot, containing quite high quality materials.

Remember I said 'in theory', it was a hypothetical. A lot of Tiny Houses are very expensive per sq ft because of the kind of people who are building them. Mine is for example because it is stuffed full of cool features, materials and designs. I believe one can build a mass production of Tiny Houses for 5k-7k per unit which have all the basic necessities.

> They are basically aimed at eccentrics

I'll take that as a compliment. I prefer 'unique individual' ;-)

> not normal people with a limited income.

You double your costs per sq ft if you hire somebody else to do everything for example.

Tiny Houses are very accessible to the lower middle class and to the working class, because they tend to have the right skillset or because they have some amount of cash flow. Most Tiny Houses are built and finished for about 25k. That is good number when contrasted with the alternatives.

If you have close to zero cash flow, you have two options available to you, perhaps to be used in concert.

1. Reclaim waste materials. People have built Tiny Houses for as little as 500 dollars although I don't recommend this if you can get a job and save up. But if your time is effectively free then this is a wise way to spend it.

2. Construct a micro sized Tiny House to be transported by bicycle. That's about 200-300 dollars. Better than being homeless with no shelter.

I think we're coming at this subject from different premises. Mine is 'obtain shelter' and yours is 'obtain housing'. I think government related problems aren't going to be solved for a long time and that in light of that a lot of people are going to get screwed over unless they adopt something like Tiny Houses.

I think we're coming at this subject from different premises. Mine is 'obtain shelter' and yours is 'obtain housing'.

Yup, that's why in my original comment I talked about affordable housing, not affordable shelter.

Good for you that you like your current lifestyle and are a fan of tiny houses. It doesn't solve the problem space I am suggesting for the question posited.

How are tiny houses better than small apartments? Being able to vertically stack things sounds hugely more efficient.
Density or going 3D does make things cheaper.

However there are still three problems.

0. Zoning in the UK and USA is prohibiting new construction. This is a fact that can't be got around. It is a political problem.

1. House prices. Extant legal construction builds have gone way above what millennials can afford on their stagnant wages. It is a logical consequence of monetary policy at zero % interest rates can cheap debt but many people, including myself, have no desire to be buying a building which will take several decades of work to pay for. Renting isn't a solution either.

It used to take between 5 and 10 years to pay off a mortgage. Builders I've talked to say their new builds in the US are never less than 350k per unit.

2. Work is not what it used to be. Even jobs which have decent wages have not kept up with the kind of house price inflation that exists in London and San Francisco. People go where the work is, and it is becoming ridiculous to shackle yourself to a thirty year debt when the job you went there for lasts about five-seven years on average.

Summation: Tiny Houses are a good solution to a world of stagnant opportunity and a schizophrenic economy. They are about putting a backstop against getting fucked by going bust or becoming homeless to put it plainly.

A modified parking garage with simple utility hookups would be a good place to park tiny houses. If you could spring ~5k for a tiny house and $50 a month for garbage collection/water, you'd have a decent situation going. Cheaper to build and more flexible than an apartment building, but still "stackable."
Mesh networking has captured a lot of my attention lately. It isn't that new, but it isn't exactly available to people. Given that we overcome some of the existing challenges with meshes, we could have deep Internet penetrantion without having to deploy a lot of infrastructure.

There are some communities on the net who are working on homebrew cybernetics. I've seen experimental designs for transdermal data/electric ports, health monitoring implants, and implanted secure elements, &c. CyborgNest recently started taking pre-orders for a "north sense" implant[1]. Definitely my favorite fringe.

[1]: https://news.ycombinator.com/item?id=11778249

What about Cyc?

https://en.wikipedia.org/wiki/Cyc

I think it could work, and best if paired with an application project. Not everything is scalable and a long, hard march may be required to create some kinds of systems, especially with respect to Agent based computing.

Personally I think Agent based computation is the next big thing after Social, but it is hard to tell what form it may take. I am far from convinced that Apple or Google have the chops to do what is necessary, because it would surely require them to violate their own interests and probably legal obligations.

Talking of that sort of thing, you should look into Urbit.

Urbit is a network computer that could hypothetically allow Agent based computation to actually work.

https://urbit.org/

Urbit is definitely fringe science stuff, I'm sure Walter would approve.

I did research in Multi-Agent Systems and always believed that they are going to be the next big thing but in the end realised that our technological capabilities are not there yet. The urbit is probably a good idea towards that kind of computation, still a long way before that.
You could very well be right, it could take decades.

I think Augmented Reality will abruptly allow a lot of non-geeks to suddenly be aware of certain realities (imagine seeing a calorie counter on your food, or the fact their data flows are going off somewhere outside of their control), so I hope that will accelerate progress in an Agent based direction.

Agents will be easier to build if they are visualized, they may allow for metaphorical understandings across millions of people.

I agree that AR will allow for software to live between us. We don't have that paradigm at the moment. Yes we have a lot of devices that we use but they are all passive, nothing that we can visualise, interact with and that is active in real time like AR entities.
Polygamy, polyamory, short-term relationships throughout life, coliving communities, not settling down, minimalism, no ownership.
Sounds like life in '68?
Good point and I think tech is making us neo-hippies, getting us away from conformity (in some ways) and closer to our desired
Editing your own genome. (Note well: I am not suggesting this!)
Honestly I would look toward manufacturing, particularly if you can incorporate ideas from the world of the digital -- for example crowdsourcing, mass customization, or continuous delivery of updates/modifications. I read an article in WIRED oh, I guess about five years ago now, called "Atoms are the new Bits" that talked about companies using 3D printing, crowdsourcing, etc, but to make real world products.

Part of the reason I think this area is hot is that it's harder to do than apps. Every "app" I can think of has already been done, and in fact there's probably a dozen versions of it languishing in app stores. Manufacturing is harder, and is often geographically local, so there are fewer competitors and more niches.

Another leading indicator for me is the general rebellion of young people against the fact that "shop class" has been taken out of schools. Lots of kids are trying to find ways to be "makers". In my state, when I was in school the "dummies" were pushed toward taking vocational classes in plumbing, welding, etc. Now (20+ years later) the vocational classes are the most competitive, prestigious classes that students fight to get admitted to.

If I had to guess, the new manufacturing is going to be smaller scale, extremely high quality, locally sourced materials, with some kind of digital "twist" on the business models.

VR, AR, and AI (not ready for strong AI yet, IMO)
Aren't these mainstream areas of research though?