arnioxux

Wasn't google tango their solution? (but requires custom android hardware for depth sensor)

https://get.google.com/tango/

I haven't heard anything about them for a few years now.

Visual-inertial SLAM has been working for a few years on mobile phones: https://www.youtube.com/watch?v=e7bjsIqlbS0 So there should be plenty of third-party libraries you can use.

No opinion on "The Man Who Knew" but kind of bummed that it will now beat "The Man Who Knew Infinity: A Life of the Genius Ramanujan" in autocompletion.

Taking "technology" to its logical end, you will probably arrive at AGI. And despite being decades off away from it, there are already many people making sure that it will be "safe" since the default otherwise will most likely lead to human extinction.

https://waitbutwhy.com/2015/01/artificial-intelligence-revol...

https://waitbutwhy.com/2017/04/neuralink.html

EDIT: I am getting a lot of downvotes for this comment. And just to clear up intention, all I am trying to say here is that yes, technology could be deleterious but we can prevent it in ways other than concluding “and that’s why we have to start blowing shit up and killing people”. I don't see why that's offensive.

Which just means "people similar to you who like x also like y".

This does make a lot more sense than analyzing the audio of the music IMO. For example youtube does this okay and if you look for a Mazzy Star song after watching Ricky and Morty (a tv show), it will recommend other Ricky and Morty soundtracks even if the style is completely different. This isn't something you can predict with just audio data.

Is the end goal to allow people to donate computing power for training? (a la Folding@home or SETI@home except just by visiting a webpage)

If so that's amazingly clever!

I would be so paranoid of leaking information if I had to do twitch live coding. For example how do you handle copying and pasting api secrets? Or just pure privacy stuff like not showing other domains you own in namecheap (which he doesn't seem to care about in: https://youtu.be/Tq1uoC-lxy8?t=528)

They also did a very thorough AMA on reddit recently: https://www.reddit.com/r/IAmA/comments/6pa11p/we_spent_5_yea...

The "JAPANESE ONLY" sign was probably meant for chinese tourists. The kanji below it, "謝绝中国人", translates roughly to "decline chinese people" which is understandable given their notorious behavior abroad:

http://shanghaiist.com/2016/12/30/shanghaiist_2016_embarrass...

https://www.nytimes.com/2016/10/11/world/asia/bad-manners-ch...

http://www.japantimes.co.jp/opinion/2016/10/07/commentary/wo...

There are plenty of adversarial examples for humans too: http://i.imgur.com/mOTHgnf.jpg

I didn't look at the code, but canvas only supports 3x3 transforms and you need 4x4 to do perspective: https://developer.mozilla.org/en-US/docs/Web/API/CanvasRende...

So if those cubes are textured, you'd need to subdivide the texture and approximate each with affine transforms. Here's an example of this hack (which was only necessary years ago before webgl existed): https://acko.net/blog/projective-texturing-with-canvas/

Can someone do a feature comparison between this and three.js?

The term for this effective altruism strategy is called https://en.wikipedia.org/wiki/Earning_to_give

I think this is the correct way to model it. In particular you can model it as a markov chain (commonly studied in queueing theory):

https://en.wikipedia.org/wiki/Birth%E2%80%93death_process#/m...

That's clever! I updated the code and it seems to work for the several times I refreshed now. Thanks!

This was the final cost function that worked:

  let encloseCost = 0;  
  let farthestPointCost = 0;  
  for (let i = 0; i < B.length; i++) {
    const b = B[i];
    const dx = b.x - a.x;
    const dy = b.y - a.y;
    encloseCost += Math.max(
      0,
      -a.r + (b.r + Math.sqrt(dx * dx + dy * dy))
    );
    farthestPointCost = Math.max(
      farthestPointCost,
      b.r + Math.sqrt(dx * dx + dy * dy)
    );
  }
  const radiusCost = Math.abs(a.r);
  
  const totalCost = encloseCost + farthestPointCost + radiusCost;

To answer my own question, the problem seems to be that once it finds an enclosing circle it loses gradient on x and y so it doesn't know where to reposition the circle. Not sure how to fix this.

I wanted to try a more naive/slow/sledge-hammer approach of just setting up the cost function and throwing it at an optimizer.

I couldn't get it to converge and I am not sure if it's just the library I am using being buggy (numeric.js) or my cost function is just bad.

Cost:

    const cost = (a, b) => {
      // If a encloses b, returns 0, otherwise a positive cost
      const dx = b.x - a.x;
      const dy = b.y - a.y;
      return Math.max(
        0,
        -a.r + (b.r + Math.sqrt(dx * dx + dy * dy))
      );
    }
    let iter = 0;
    const costAll = ([x, y, r]) => {
      // Cost of enclosing all balls (which should be 0) plus cost of radius (which should get us the smallest ball)
      const a = { x, y, r };
      const radiusCost = Math.abs(a.r);
      let encloseCost = 0;
      for (let i = 0; i < B.length; i++) {
        encloseCost += cost(a, B[i]);
      }  
      const totalCost = encloseCost + radiusCost;
      return totalCost;
    }

Demo:

https://codepen.io/anon/pen/ZyEaaQ?editors=0010

EDIT: demo should work correctly after mbostock's hint below!

http://projectnaptha.com/ is also a very similar project. I think it uses Ocrad.

The creator did an AMA on reddit: https://www.reddit.com/r/StardewValley/comments/4b8s7v/im_co...

I thought it was pretty inspirational that he managed to keep at it for 4 years solo even when his grandma and himself started to doubt that he would ever ship.

Exact link: https://news.ycombinator.com/item?id=35079

I guess no one managed to top that comeback in the last decade (thread is 3600 days old!).

OP's hyperbole took it too far by saying "good programmer". But if you change it just "top one percent programmer", it might not be too much of an exaggeration. Before I also get banned, I want to clarify that there's a difference between:

- "it's hard to find a woman programmer in the top 1%" because women can't be good programmers due to their biology

- "it's hard to find a woman programmer in the top 1%" because our society failed to train and produce them

The second statement isn't inherently sexist, it's just an observation of the poor state of our world.

If you're familiar with programming/math competitions you'll know that there are not many female represented there even after taking into account their smaller population: http://www.topcoder.com/tc?module=AlgoRank. There are probably 500 "red" coders (which is roughly the top 1%) but with probably less than a handful who are female. All you need is a username to compete so it's pretty objective and gender-blinded.

So it seems exceptional female programmers are simply not being produced even as early as high school and college level. Google interviews are very similar to those competition so I assume their data will also uncomfortably show this.

I'm not a scala dev but I'm curious, what makes scala a good scripting language? Bash has its faults but it's still great for adhoc stuff and when it stops being great you can just move to a more specialized build config tool. Scala seems a bit more verbose and it's unclear what you gain from it.

Is it similar to what's going on the JS world where people just want to use one language for client, server, and build tools? (and for this case, even the terminal shell?)

A lot of components library call these steppers:

http://www.material-ui.com/#/components/stepper

https://react.semantic-ui.com/elements/step

https://ant.design/components/steps/

I was trying to pick a react components library (for fast prototyping) and one of my requirements was having this particular component.

In at least two other companies I've worked at we also use query params to enable debug information on live production sites. At one of those companies the only requirement was that you be on a corporate ip address but it actually still works if you're on our guest wifi.