marshallp

Google Research and Stanford researchers test out a 1 billion connection, 9 layer, 16,000 core deep learning neural network (see geoff hinton and andrew ng talks on youtube) to recognize 20,000 different objects in images (with low accuracy but huge improvement over previous approaches).

This was done no with no pre-labeled images (except for fine tuning)! A brain that learned from raw images. The same algorithm can be applied to any data type (financial data, text, audio, images/video) without any human involvement (except gatheting of unlabelled data and running the system). Pretty much the artificial intelligence holy grail!

I agree, the classic engineers like to work problems with an easily definable end result (so that business types don't tell them what to write).

It's a pity though, there's a huge space of products that are needed but the talent is going into slugging it out in societally useless and relatively low paying niches (such as many of the most talented programmers entering the game industry or social networking or linux kernels). Meanwhile dumbasses are raking in millions with shit like SAP and the like.

The majority of amazing software that has not been created yet is "CRUD". Why developers keep thinking the low level stuff is the good stuff.

A huge class of such systems that are waiting to be created are crowd labor systems. How do you create leaderless organizations that produce what google or boeing or walmart produces and that compensates individuals for performance.

The answer today is a messy unorganized market system. Proper information systems that better organize that would create a lot of value. The analogy would be this site. Thousands of people communicating, without the organizing software it would be much worse (thousands of people standing around talking to each other, the good ideas spoken would not easily reach the top as it does in this organized software system).

tldr; the biggest value opportunities lie in CRUD apps, not the low level code

How about the best of both worlds (and more) - ocaml/haskell

Seeking work - remote - toronto, canada

Favorite technologies - haskell,python,postgres Also do javascript,java,lisp,php,mysql,appengine,android

can develop webapps/mobile apps from scratch with just loose requirements

i particularly enjoy writing tests and doing maintenance on codebases

also enjoy develops, automating systems

also big frontend work in gwt

also can build scalable systems on appengine

solv-able, not solv-ed yet.

So you're saying people should give up on the idea of living to a millions years old and just suck it in by spending more time with the families/friends in the few years they have?

Death is a solvable problem. There are species and cell types that are essentially immortal. It exists in nature. We have the capability to manipulate things at the cell/molecule level.

I'm not sure what you mean by "colonized by industrialization". If you think things are too industrialized, maybe you should give up the laptop you're writing from and go live in the jungle without any plastics r metal. I'm sure after a day or so you'll come to realize the error of your thoughts.

Actually, these aren't "hard" problems in the sense that physics disallows them. We have existence proofs of them. It's is just a search problem. You have to set up robotic systems that search the space. So the problem is really one of collective stupidity ie. most people are dumbasses debating they value of time and spending it family/friends, or whether gays should be married, when instead they should be demanding mass industrialization of scientific discovery, or building they it (surplus labor is instead going to youtube/wikipedia/tv watching/novel reading etc.).

Most people think science is done with eureka moments by geniuses, but actually what is needed is a "systematic" exploration.

Humans in the heat of the moment do stupid violent things. I would bet weaponized drones would cause less fatalities than in person officers. Make the decisions completely autonomously by computer and you will probably see even less mistakes. Plus, there is a huge class of crime, the inner city, where police do practically nothing currently. Drones might well stop gang violence.

The possibilities really become clear though internationally. If foreign governments, such as mexico, recruit drones, they might well solve their violence problems pretty quick.

It just takes one succesful lawsuit by a user.

Azure SQL is only $10 per month for 1GB. Why is heroku so ridiculously more expensive than them.

If I were them, I would just run the vehicles as a fleet. The have many billions in cash and borrow any amount, so why not. Then innovate on the maintenance of the fleet with robots as well. Maybe start with franchising. Plus, they automatically set the rates for each individual journey dynamically and extract as much as possible.

I'll plead WTF on that.

I'm simply trying to tell the kids on here that SQL is great and ROR/LISP/NoSQL worship around these parts is misguided. No 'droppings' in that.

Yes, I write Lisp on my resume to make me look like a stronger candidate. I learnt it a few years ago for fun, that's all (I'm not a lisp flip-flopper). Just want to put the advice out there to be a relational weenie (instead of lisp weenie).

They're not the same. Unions are equivalent to crony capitalism (ie. regulations put in place to prevent competition), but most of what companies do is just free markets at work (which helps the consumer).

Microsoft made bank with the kinect. Also, google is in serious danger of being killed by facebook (more data = better search results), so google needs another huge profit stream. Running the entire transportation infrastructure can be that (there will be competitors, but with google's superior tech talent, their vehicles will be safer and hence win).

How come you're not using matlab/numpy?

Why WTF! I'm looking for AI type work, which is much much rarer than sql work, which was my point. You wouldn't use lisp for that either (matlab/numpy followed by c optimizations is used there).

If 'components' were that useful for the things rdbms's are used for the thousands of highly trained phd's at oracle, microsoft, ibm would have added them to their sql products.

Of course, this a ROR crowd, software invented by a business school grad/game review writer, so it's can be an uphill battle explaining this stuff.

I've explained a bit longer than your 'wtf'. Perhaps you can explain to me applications where lisp is the best fit?

Almost every website or software running a business I can think of runs sql. The very small few that don't, google for example, run on c++ and assembly.

So, in 2012, with gigs of ram and postgres, why would I use lisp?

Not really. Obama's foreign policy is better than the republicans and he genuinely cares about people so healthcare and welfare policies would be better, but his support base also consists of the unions, which I disagree with.

So basically a thin sliver of what an sql rdbms provides.

Well, basically, most applications that you want to write are basically 90 percent rdbms with a thin layer on top for the views (usually php etc.). The 'power' of lisp isn't relevant. Other types of applications are usually performance critical and written in c etc, or glue code written in languages with libraries tcl/perl/python etc.

So basically the traditional niche of lisp, complex apps, has been replaced by sql.

There are apps written in lisp, like this site for example, but that's just for vanity mainly (pg is famous for his lisp books) - it would be have been easier to write it in sql.

There's been talk recently of wealthy americans fleeing the country for tax reasons. If Obama gets reelected, as is expected, he's gonna go all out in his second term (I'm an Obama supporter but I dislike the idea of taxing the entrepreneurs more to fund the lazy government workers).

Why not just do relational programming - prolog or sql. The lisp weenies still don't get it.

Of campaigns, basically a scrape of their website (they make their money fees, not advertising). Then other people can do review sites of campaigns and do statistical analysis.

The biggest new technology breakthrough can be computer vision (so that unconstrained robots are possible). However, the type of people who can work on it are mostly working in big data (the stock market and online advertising).

It is reasonable for them to do so (the money is just sitting there to be taken), so the solution to the innovation problem is to fund more people to work on computer vision.

This is why kickstarter/indiegogo need to release an api or a data dump. It will only help them and the public, just the same as the sec releases public company data.

Or "developer". You can probably plot the demise of programming as a respected profession as their titles changes from programmers to "software developers" or "architects", maybe using google books n-gram.

How about flying carpets (Neil Gershenfeld at MIT is working on making one)

It is already done to an extent (high-throuhput machines), however, there's still a lot of old timers in biology who spent too long pipetting and haven't invested yet.