Disqus - Latest Comments for crines

Re: Hierarchical value functions in deep reinforcement learning

Chris — Mon, 22 Aug 2016 00:17:06 -0000

Hi there, even with the open issues this addresses a number of issues in other DQN Models making it a VERY interesting piece of research - congratulations.

I am wondering if you have released or have available a reference important Implementation?

Cheers,

Christopher

Re: Hacker Poll: Do You Plan to Learn Any New Programming Languages in 2011?

Chris — Wed, 29 Dec 2010 19:40:20 -0000

I've been slowly learning CoffeScript but I'm going into overdrive and will be digging deep in 2011 so the answer is I know a bit about it now but not enough to consider it a core proficiency so I'll b "learning" more in 2011.

I missed your node.js clarification. Still not sure it's a language but I do think I understand your reasoning for inclusion.

CoffeeScript will be huge, I'd investigate it for future articles. :)

Re: Hacker Poll: Do You Plan to Learn Any New Programming Languages in 2011?

Chris — Tue, 28 Dec 2010 23:29:44 -0000

Klint,

you're missing next years big one, CoffeeScript. also I don't consider node.js a programming language!

Christopher

Re: We need an independent invention defense to minimize the damage of aggressive patent trolls

Chris — Tue, 19 Jan 2010 15:42:07 -0000

We have quite the stream going here. :)

1. My concept does not apply to today's broken, abused and abusive patent system but to how I would envision it work if we took away the concepts of Generality AND the ability to patent the outcome of a State Change which is where most of these silly patents live. So in your example today that would work (UG, Yuck, With You it Sucks) but in my revised system it would not.

2. The concept of independent invention could be applied to these strict, novel, provable algorithms and should not be applied to generalities (they should go away all together).

3. Ok here's my take on the origins of patent law and how it was intended to protect people vis-a-vis the Constitutional rights of freedom. My thought is simple but I believe valid. Patents originally existed so 1 person could create something unique and not be run over by a huge company or persons with deep pockets. This is aligned with your protection.

4. Patents in the world of software are abused! The concept of patenting the outcome of an algorithm is INSANE, WRONG, BAD because as we've both stated I can get the exact same output in an unlimited number of ways. This too aligns with the original spirit of patents, ie. you could patent the specific method to make toast but not toast itself. Another could build a toaster to make the same toast a different way or they could improve upon your toasted patent without infringing on your toast making "algorithm".

So in conclusion I believe a reformed patent system (see all my above comments) with checks and balances gets back to the original spirit of the law which is protecting individuals right to create something and be fairly compensated for it. Patents started as pro little guy but in the last 20 yrs or so (as you state) have been utterly abused, if we get back to the original premise using physical machine laws as the bases (as my idea basically does) and have Independent Invention Protection of some sort for times 2 independent camps create an identical algorithm (not outcome but mathematically provable as identical) it MAY work again.

Re: We need an independent invention defense to minimize the damage of aggressive patent trolls

Chris — Sun, 17 Jan 2010 23:30:27 -0000

I don't think what I wrote was confusing but I guess it was...

It's not actually, a novel algorithm is provable as such.

Additionally 1 patent can NOT cover boundless specific methods, in my example only 1 specific, provable and novel algorithm can live in a patent.

What I am suggesting removes the concept of a generic patent. So for instance in my example the Bayes' theorem could be patented but not how or what it is used on. If another like method was created which was mathematically different (but could even have the same output) then it would be valid. Both algorithms could be used to create a spam filter with no patent infringement on the spam filter idea itself as it is the outcome of an algorithm not an algorithm in an and of itself.

Please explain to me where in the constitution the concept of progressing science and arts is covered? This is patent law which as far as I (a non-american) understand is not codified in the US's constitution. If I am incorrect I apologize but otherwise let's keep to facts.

Finally ideally the world would be patent free and execution would rule but even though I hate patents in today's world I can understand the need for the protection they provide.

If they could be abolished I'd be a happy camper but that's just a little too perfect world to be reality. So the best bet is only allow a strict method to be patented just like the original patents did.

Re: We need an independent invention defense to minimize the damage of aggressive patent trolls

Chris — Tue, 12 Jan 2010 04:06:54 -0000

It's very simple actually... The method of state change (in software's case this is an algorithm) can be patented, not the state change itself but the method by which the state change occurs. There are an unbounded number of ways to change state A to state B and each has the potential if they are TRULY novel to be patented.

This is in keeping with the original nature of patents. For example the method used to turn bread into toast is patentable BUT NOT the toast itself!

Re: A Day Without Disqus

Chris — Wed, 06 Jan 2010 07:38:24 -0000

Hi David,

that's not quite correct... The saying is people eat with their eyes but taste is still the most important thing and I think that holds true for comment systems as well.

Re: The next big thing will start out looking like a toy

Chris — Mon, 04 Jan 2010 18:38:42 -0000

I misunderstood and stand corrected. :)

You are saying that incumbents and possibly the public viewed as actually disruptive services/products as "toys". This I very much agree with (and have experienced both sides first hand).

Cheers,

Christopher

Re: The next big thing will start out looking like a toy

Chris — Sun, 03 Jan 2010 17:32:06 -0000

Hi Chris,

Generally I agree (9 times out of 10) with your startup thoughts, especially given you and members of your super team (like ex-vancouverite Catrina) have done it before (and I've just been building for past 8 years) but I have a couple of quibbles on this post (but not necessarily "disagreement").

Flickr is a perfect example of your thesis so I see where the germ of this idea may have come from but I think you're missing that most toy startups remain toys. While startups who strive for non-toys (Google, Microsoft, Amazon, etc) tend to be the real big winners.

I also think (bring on the hate mail) that this type of thinking, also espoused by YCombinator is hamstringing many startups.

So what do I think will be big in the next 10 years? Startups focusing on non-toy solutions for Big Data, Intelligence & Goverment, Publishing and New Device Optimized User Interfaces.

Cheers,

Christopher

Re: rssCloud news (Scripting News)

Chris — Fri, 07 Aug 2009 01:51:18 -0000

A lot like FeedTree from a number of years ago...

FeedTree.net

Re: Some Notes on Distributed Key Stores

Chris — Tue, 19 May 2009 00:28:33 -0000

Great write ip, chocked full of good insights.

I know the pain of having to build my own fundamental library as what existed is just not quite what I needed.

On the topic of key/value stores you missed one which is a little more obscure but I like it a lot. SkipDB from the author of IO.

Re: Josh is right, URL shorteners are risky (Scripting News)

Chris — Fri, 03 Apr 2009 21:14:57 -0000

I have no issue with some URL shortening programs but what I'd like to see is an automated way to store the actual url in a bookmark service of our choice, both as safety & a bit more findability.

Cheers.

Re: That's Only Ten Lines Of Code

Chris — Mon, 30 Mar 2009 22:39:03 -0000

"But just because something is simple to build doesn't mean it won't be a good venture capital investment. I've heard many people say they could build twitter in a weekend. I don't doubt it. But twitter is more than code"

People who name the classic mistake of equating a complex system to something that is preferable really irk me. I love working on difficult problems and coming up with cool solutions and some things do require a lot of deep coding and complex algorithms to solve BUT if something can be solved in a simpler way or doesn't require a complex solution why does that devalue the solution. Well it doesn't.

Re: Portable Contacts lets you get friends’ info on other sites

Chris — Fri, 27 Mar 2009 00:57:02 -0000

OAuth + Portable Contacts + Open Social Graphs = The next big thing in Social Networking, closed networks better get it before it's too late. Add in a bit of mobile location data and things start to get exciting.

Re: http://www.avc.com/a_vc/2009/02/how-patent-trolls-are-a-tax-on-innovation.html

Chris — Sun, 15 Feb 2009 14:28:36 -0000

One of the biggest issues with patents as they stand today is we protect the outcome rather than the method used to create the outcome.

In software there is an infinite number of ways to get to the same outcome so if we must have software patents let's protect the underlying algorithm not the outcome of the algorithm.

Re: Introducing Two-Click Signup, an initiative to improve the user experience of OpenID; first test now live with Google

Chris — Sat, 31 Jan 2009 02:49:12 -0000

Question: Do you see the hybrid OpenID + OAuth as the best method for implementing this or is it a stop gap until a new workflow is created?

Thank You.

Re: Introducing Two-Click Signup, an initiative to improve the user experience of OpenID; first test now live with Google

Chris — Fri, 30 Jan 2009 03:33:17 -0000

Looks like in the Google Blog about this there are some libraries to test this out in other client implementations.

Re: Introducing Two-Click Signup, an initiative to improve the user experience of OpenID; first test now live with Google

Chris — Fri, 30 Jan 2009 00:26:15 -0000

This is really cool & exciting, the next big step is to allow the broader community to set-up similar connections with Google.