tst

Atul Gawande wrote about 'non-human' tools in medicine [0]. He advocated using checklists but he has also written about other tools. One I remember was a flow chart to predict some sickness. It outperformed humans solidly.

The biggest problem with checklists, flow charts and software (one older example is MYCIN [1]) is adaption. If the medical practitioners don't want the systems they are going to fail. He also writes a lot about that in the book. He tried to introduce it into hospitals but the professionals often ignored the lists or just checked the boxes without actually checking the condition.

I personally imagine that a practitioner with a good system works the best or like Frederick P. Brooks said [2]:

> If indeed our objective is to build computer systems that solve very challenging problems, my thesis is that IA > AI that is, that intelligence amplifying systems can, at any given level of available systems technology, beat AI systems. That is, a machine and a mind can beat a mind-imitating machine working by itself.

[0]: http://en.wikipedia.org/wiki/The_Checklist_Manifesto

[1]: http://en.wikipedia.org/wiki/Mycin

[2]: http://www.cs.unc.edu/~brooks/Toolsmith-CACM.pdf

I'm also recovering from a depression which lasted for quite a while. It absolutely sucks because you think you're worthless, nobody loves you, you can't get anything right and the best would be if you just wouldn't exist anymore.

And on top of that you isolate yourself. I know how hard it was to ask for help therefore I want to show you some things which helped me:

- Realize that your depression is lying to you. It doesn't tell the truth. It makes you believe that something is logical even if it isn't.

- Read 'Feeling Good' - terrible title, great book. It will probably work better than average on the average HN reader because it takes a 'rational' approach to depression (cognitive-behavioral therapy). It helps you to recognize destructive thought patterns and how to deal with them.

- Garbage in, garbage out. What works for computers also works for your body. Yeah, you're a geek but you can eat some vegs instead of the 500th pizza. Also working out (or other sports) are pretty great.

- Long term: Therapy which tries to work on the root cause and not just at symptoms.

Finally, here's a rather extensive list with lectures, books, exercises, etc. which help dealing with depression [1]. Back when I was fed up with feeling crap I created a spreadsheet with the 8 activities and tracked those every day.

Note: Every person seem to react to differently. I read about people who improved a lot by meditating - on the other hand, it didn't work for me.

So, try some things out and don't give up. You can beat that liar in your head.

[0]: http://www.amazon.com/Feeling-Good-The-Mood-Therapy/dp/03808...

[1]: http://www.reddit.com/r/getting_over_it/comments/1nd14u/the_...

PS: If you have any questions feel free to ask - if you want to send me a private one write at <username> @ panictank.net

I think the analogy to Wozniak is great. I remember an article some months ago in which he basically said the same: He wants to tinker and hack.

I remember watching some parts of notch's livestreams. I loved the enthusiasm he had. He was a bit like a young boy, trying things, throwing some away, creating games. I'm happy for his decision.

About 7 years ago I wrote my first line of Python and I loved it. Before that I had some experience with C and PHP and the clarity of code and the great documentation was mind blowing.

The slogan 'batteries included' was actually true. You could do so much stuff out of the box and I learned to love the language. The community back then was pretty small (at least the German-speaking one) and they welcomed new people.

In the last months it doesn't feel right anymore. The hate in the community discourages me. It doesn't feel like the Python I learned to love anymore.

And I'm sad about that development.

About 3 months ago I looked around for an other language. I wasn't up-to-date on the language development since Haskell became more popular (around 2008?) and quite a bit happened since then.

So I looked around, looked at some code, some new languages, read a few blog posts on each and finally settled on learning Clojure. I loved Lisps (more specifically Scheme) but the environment was – imho – a bit problematic.

I adopted a lot from functional languages into my Python coding over the years so the transition wasn't that hard. And I really love the language so far. The community seems to be quite active and is welcoming. :)

You can check out Machine Learning: An Algorithmic Perspective by Stephen Marsland[0] which takes a less math-driven approach to ML. The code is available online if you want to take a look at it (it's written in Python)[1].

[0]: http://www.crcpress.com/product/isbn/9781420067187

[1]: http://seat.massey.ac.nz/personal/s.r.marsland/MLBook.html

I can recommend scrapy[0] if you work on a bit bigger problem. But even then if you familiar with scrapy it's incredible fast to write a simple scraper with your data neatly exported in .json.

[0]: http://scrapy.org/

I can give you some feedback if you want. In the past I have searched freelancers in the web, app and design space. For each position I maybe look at ~200 freelancers. That means a freelancer is out on average in maybe 10 - 20 seconds.

What I like:

* You have a website

* Design is very easy and clean

* You show your references

* There's an e-mail address

What could be better:

* Your picture. If you don't want to show your face just leave it out. Otherwise take a nice picture and put it up without any effects / filters

* I don't know what you did on each of your projects. Normally, I assume that you either did the whole site or a minimal part.

Stuff that depends (if you mainly sell to technical or non-technical people):

* Your services part doesn't really describe technologies / frameworks. However, if I look for a freelancer I look for somebody knowing Django, Magento or whatever.

* Same goes for the projects

As someone with no experience in computer graphics it's always insane to see demos like this. Especially using just around 250 lines of javascript. Really impressive.

Also on the topic, the demo scene stuff is mind blowing, too. [0]

[0]: http://awards.scene.org/awards.php

Here's a pretty good talk about restricted Boltzmann machines by Geoffrey Hinton. He explains the concepts and problems very well and basically without maths:

https://www.youtube.com/watch?v=AyzOUbkUf3M

> I believe in the real world, you are best represented by a body of work and/or the ability to demonstrate that you are capable of tackling the problems associated with the job in which you are considering.

This is such a good advice! I remember reading Norvig's spellchecker[0] and said to myself: "This code is so elegant and beautiful. I could never write something like this."

I learned more about NLP, learned some Scheme, worked through Norvig's Design of Computer Programs[1] and improved my Python a lot.

What have I learned? Firstly, people like Norvig have tons of experience and are incredible clever. Also they have worked in the same (or similar) domain for decades. Of course they know cool data structures and algorithms for NLP problems. Back then I didn't realize this and thought that I was dumb. Today, I'm less dumb and know that it takes time and experience. Build, learn, build, learn, ...

[0]: http://norvig.com/spell-correct.html

[1]: https://www.udacity.com/course/cs212

I would even go further. The biggest problem isn't just using p values and R^2. I think the biggest problem is that a lot of people didn't learn statistics properly.

Properly is a vague term. So what do I mean? Instead of obsessing with tons of techniques going back to the basic and actually learn how to do design studies, work with data, learn statistical reasoning and critical thinking.

I took quite a few courses in statistics because I liked it. But a lot other people – especially those that apply statistics – maybe take one or two courses in stats and then do research / studies. The results can be pretty terrible. In conclusion, more fundamentals and less icing on the cake.

I completely agree on the package manager issue. I switched to OS X about 6 years ago. Before that I mainly used Arch Linux with wmii. My switch from Arch to OS X took about 6 months in which I slowly migrated. At first, it was pretty great but when I wanted to upgrade software or install some new one it was horrible. Back then I used macports and I can't remember all the problems. It was a downgrade from pacman (and apt). I gave up coding a bit because it wasn't so much fun anymore.

Lately, I started using a linux vps as my development environment and it's so much better. It's worth switching just for a good package management system.

I would be interested to see how this effect affects the conversion rate. I could imagine that it could work very well on physical products.

Just a small tip which may ease the search for methods. The general term for "on the fly" learning is online learning [1]. The rest depends on your problem but there are often online variants of offline methods, e.g. when you work with Gaussian process regressions

[1]: http://en.wikipedia.org/wiki/Online_machine_learning

I think these numbers are based on one blog post the company made[1].

They wrote:

"[...] defined that an app has to hold a position for at least 7 days to be considered as "ranked". That was the case for 265,959 apps from the 18th of July 2012 to 25th of July 2012. To the remaining 410.023 apps off the ranks, we refer as app zombies, leading a life outside a prospering market."

and

"[...] in theory app zombies can still be downloaded, we concluded that an average zombie is getting zero to ten downloads a day, depending on the country."

[1]: http://www.apptrace.com/blog/2012-08-06/inside-zombie-land

There was a site called hearablog.com (https://twitter.com/#!/hearablog) which narrated some of the more famous tech blogs. Jason discussed it on This Week in Startups: http://www.youtube.com/watch?v=gTn5qNeHny4

I personally think that this is a neat idea but VoiceBunny isn't the ideal tool. You pay about $70 for 400 words which is about the length of a short-mid sized blog post. If you just cover 5 posts per day, you will need to invest about $10k per month in VoiceBunny. At this point, looking for your own voice actors is probably cheaper.

The data is pretty easily accessible as JSON - however, here's the data I used: http://pastebin.com/nRYv40U8

Firstly, a boxplot with the quotient of entertainment contributions to entertainment & internet contributions.

http://i.imgur.com/FWQWy.png

You can see quite easily that there's a difference which is also significant (95%, t = -4.73).

I've also done a logistic regression correcting with age, party (is_democrat), seniority and quota of contributions (quota_ent).

  ------------------------------------------------------------------------------
       support |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]  
  -------------+----------------------------------------------------------------
           age |   .0258551   .0358136     0.72   0.470    -.0443382    .0960485
   is_democrat |  -1.252883   .6243361    -2.01   0.045    -2.476559   -.0292067
     seniority |  -.0262688   .0381962    -0.69   0.492     -.101132    .0485943
     quota_ent |   5.839435   1.447732     4.03   0.000     3.001933    8.676938
         _cons |  -1.968467    2.01512    -0.98   0.329    -5.918029    1.981096
  ------------------------------------------------------------------------------

The AUC is 0.8089 which is quite okay. Furthermore, it would be interesting to test whether location is a significant factor.

Edit: @adamtaylor: Here's a scatter plot with each contribution, transformed with log(1 + x) for readability: http://i.imgur.com/MRciL.png

Here's the full report at Science: http://www.sciencemag.org/content/334/6061/1413.full

Yes: http://thenextweb.com/eu/2011/01/13/german-google-analytics-...

However, I've never heard of someone paying this fine.

Probably Kuka: http://drstienecker.com/tech-332/11-the-kuka-robot-programmi...

Reminds me of the infamous Rapidshare Captcha with cats. [1] There were hard to solve. However, also for humans.

[1]: http://www.labnol.org/internet/favorites/cats-inside-rapidsh...

That doesn't mean so much if you don't know the effect size and variance. In the end you have to look at the confidence intervals.

Under the law of demand, i.e. rising prices decrease demand, it is overpriced when lowering the price increases total revenue. Reading the comments on his post, it could be overpriced.

There's a really good book about this (sadly only in German). However, the author[1] writes about his first business which was basically a mail order company for Darjeeling tea. He imports the tea, a company packs and ships it. Almost all other parts (accounting, customer service, etc.) are also outsourced. In 2005, the company had a sales volume of about $10m.

[1]: http://en.wikipedia.org/wiki/G%C3%BCnter_Faltin

The comment by Myzplk is quite insightful:

They're profit margin is impressive, but there are quite a few companies in the same range - http://ycharts.com/calculations/rankings/sectors/Technology/...

Which part of Finance? Banks? Trade platforms? Funds?

Also there's an older thread on HN: http://news.ycombinator.com/item?id=568556

Depends on your product and customers.

http://news.ycombinator.com/item?id=1907332

http://news.ycombinator.com/item?id=335514

http://news.ycombinator.com/item?id=1787533

http://news.ycombinator.com/item?id=302202

http://news.ycombinator.com/item?id=375417

You're welcome! I think I see your problem in 3, (2 isn't too hard, you can buy your DNA sequence for ~$200). There are a lot of logistics companies which may have valuable data but can't mine them properly (especially the smaller, not Fedex or DHL).

I don't know which industries interests you, but there are some industry sectors like logistics, agriculture or finance which heavily rely on new insights from their data.

You could start as a consultant for one company and build a product (consultingware). If it works successfully offer it other companies (http://nukemanbill.blogspot.com/2008/06/how-to-sell-your-sof...).