Disqus - Latest Comments for maddalab

Re: Experience knows a Name

maddalab — Sun, 19 Jul 2009 21:56:17 -0000

I was just reading your post on "Array Type Hints in Clojure", you start with "How to encode type hints for Java array types came up recently in conversation with a friend and I found it difficult to Google for so I decided to write it up here."

Do a semantic search or a name search?

Re: Experience knows a Name

maddalab — Sun, 19 Jul 2009 21:52:44 -0000

>> Both types of developers can use Google right (it’s a big “index” after all)?

Google does not do a good job on semantic searches and the Internet as it is today isn't http://en.wikipedia.org/wik...

Junior developers who do name searches really do not understand the semantics of the search (example language feature they are trying to use), but Google lets them be functional while they gain this understanding thru practice.

A case of the worse-is-better philosophy in practice, I think.

Re: Creating Executable Jars For Your Clojure Application

maddalab — Sun, 19 Jul 2009 21:45:28 -0000

All of what you have said about creating a jar is generally applicable to using Ant to create executable jar. Having said that I would not merge dependent jars into a single jar file. This is not a workable solution in most enterprise environment, as an example it is a poor choice for an OSGI environment like Apache Felix.

In a traditional Java environment (runtime) I would rather use the Class-Path manifest entry to specify the dependent jar.

Re: Don't use css or table layout, use Sass and Compass

maddalab — Wed, 04 Feb 2009 13:09:40 -0000

* behaves like a programming language, so you can have fun refactoring your code
Refactoring is fun in programming languages because you see the static structure of your programming change and also are able to visualize the dynamic behavior captured in the static structure. One example would be a refactoring that you undertake to make it possible to add a new feature. The semantic disconnect between what you expect and what you get via this refactoring process is minimized (via unit tests as an example) in any other language except css (if you want to call it a language). The only time I decide to delve into css files in when I would like to change the dynamic nature (data being the static nature) of the web page, a tweek here and a tweek there and voila and new page that resembles nothing like it did before and an absolute disaster (this might be just my experience tho). This is where css is broken with regards to refactoring, the distance between what is expected and what you get (Part of the problem is also the difference in implementations between browsers). I have not tried compass or Sass yet, but from your examples I do not see how either of these tools reduce the distance between what I expect and what I get (for a beginner). I see that both as marginal improvements if any over css at best for this item

* allows you to make your layout code Dry
Useful but css lets you do this too especially if you use it correctly. Again a marginal improvement

* is easy to install and start with
When using strictly css install is a non issue, and given that I think for the other cases it is only a marginal improvement, just that I have to worry about install is a drawback

* takes care of the IE 6 incompatibilities by using the code from frameworks (I also use the excellent ie7-js library from Dean Edwards)
A definite improvement in this aspect.

All said and done, I am not motivated to invest more time in another tool chain with low roi. May be I just do not get it, or you blog does not do it justice, either way, I'll pass

Re: The Abstraction Machine

maddalab — Tue, 18 Nov 2008 21:36:36 -0000

>> But what makes a neuroscientist qualified besides his understanding of how the mind works?

Does a neuroscientist really understand the mind, or just the physiology of the brain? Is it the psychologist who really understands the mind? All interesting questions with no universally acceptable answers.

To set a reference point for our discussion, abstraction = "knowingly ignoring". In computer science parlance we do not unknowingly ignore the details. We know of their existence and build higher levels of abstraction over the details to reduce complexity. I think we can agree on that.

In my example neither of the 2 men knowingly ignored the details there were just not aware of them. Imagine a scenario where the 2 men were aware of the details (time of day, weather at the time, direction the animal was coming from), they would not be able to determine a way of escaping, the level of complexity in determining the probabilities associated with the consequences of each action, derived from multiple inputs, would be insurmountable and the men would be killed before attempting a solution. Had our ancestors reacted in such a manner, neither of us would be having this conversation on abstraction.

Question: Does the mind work in the realm of epistemology using probability theory or abstraction?

Consequence: And hence my original contention "The analogy you draw between how the mind functions and programming is tenuous at best. "

>> Very true. Any ideas on making the quantum leap of "mutation" that you describe?

Not certain that every "mutation" would lead to a quantum leap. But an example of mutation relevant to this thread, for you, would be trying to consider an answer to my last question, while I do the same (Jeff Hawkins' book On Intelligence in my shopping cart, saved for later at this time).

And here is the tunnel that might lead to the light ....
Human : The science of what makes Us Unique (Michael S. Gazzaniga).
On being certain: Robert Burton

Re: The Abstraction Machine

maddalab — Tue, 18 Nov 2008 15:19:52 -0000

The analogy you draw between how the mind functions and programming is tenuous at best.

And what I say below is equally tenuous since I am far from being an expert at neuroscience

The mind does not abstract away "patterns" to predict future events, it uses memory based on experience. Here is an example similar to yours

Man sees a hungry tiger approaching, man climbs nearest tree, consequence man lives another day. The next day man sees a hungry lion approaching, man climbs nearest tree, consequence man lives another day. The third day man sees a hungry leopard approaching, man climbs nearest tree, consequence man does not live to see another day.

A bystander sees the same sequence of events, sees a hungry leopard approaching him the next day and runs, bystander does not live to see another day.

The first man was not guilty of abstraction and neither was the second of application. They were both guilty of learning from experience.

I do agree with you that "It’s no coincidence that abstraction and application are the foundation of the λ calculus" better suited to programming over imperative languages with the caveat "in certain cases"

Both techniques are a consequence of "experience" and not a function of the ability of ones brain.

To summarize "But when I introspect my own mind, abstraction and application is the closest thing in the physical world that I can relate to what I see."

Ahh but that might be a consequence of not having experienced otherwise, drawing from Genetic algorithms, what you really need is a mutation.