Profile

Cover photo
Colin McMillen
Works at Google
Attended Carnegie Mellon University
Lives in Boston
10,229 followers|3,786,389 views
AboutPostsCollectionsPhotosYouTube

Stream

Colin McMillen

Shared publicly  - 
 
Spent the long weekend with +Kristen Stubbs​ & +Kate Tsui​ & +James Dalphond​ in Guilford, CT and environs :) Lots of fun beachcombing, eating delicious food, and just chilling out.
7
Add a comment...

Colin McMillen

Shared publicly  - 
 
"""[Keynes'] essay looks at exactly this point: when will we actually be able to supply everyone’s needs with not all that much work? He thought it would be some 15 years or so from now and we’d all be working 15 hour weeks. Simply because productivity would have advanced so much that that’s all we would need to work.

And this usually brings out the people shouting about why it hasn’t happened yet. But the thing is that is has happened, just not in quite the manner that everyone thought it would."""

http://www.forbes.com/sites/timworstall/2014/07/07/googles-larry-page-on-the-40-hour-work-week-jm-keynes-got-there-first/
11
1
Scott Dier's profile photoDaniel Egnor's profile photoJustine Alexandra Roberts Tunney's profile photoSteve S's profile photo
5 comments
 
"Our working hours [...] have declined for men and they have increased for women over those decades [...]" Sounds like patriarchy. There's an alternative: https://justinetunney.com/matriarchy.html
Add a comment...

Colin McMillen

Shared publicly  - 
 
aw yiss
 
Generate your own Neural Network inspired images with DeepDream

Two weeks ago we blogged about a visualization tool designed to help us understand how neural networks work and what each layer has learned (http://goo.gl/pUfbyH). In addition to gaining some insight on how these networks carry out classification tasks, we found that this process also generated some beautiful art.

Now you can make your own images using an open source IPython notebook, which allows you to choose which layers in the network to enhance, how many iterations to apply and how far to zoom in. Alternatively, different pre-trained networks can be plugged in.

It'll be interesting to see what imagery people are able to generate. If you post images to Google+, Facebook, or Twitter, be sure to tag them with #deepdream so other researchers can check them out too.
21 comments on original post
20
8
Bridget Spitznagel's profile photoDavid Saff's profile photoLucy Hadden's profile photoCharlie Ranlett's profile photo
 
Let slip the dogs of nightmares (:
Add a comment...

Colin McMillen

Shared publicly  - 
 
A thoughtful reply to a very important question.
 
No. It's not. 

The way the United States went after Islamic extremism was a moral and practical disaster: the government lashed out almost at random, imprisoning Muslims virtually at random, attacking countries with no connection to the problem at hand, and instituting mass surveillance programs that produced virtually no results. The orgy of misplaced state violence which occurred in the aftermath of 9/11 was a powerful statement against Islamic extremism -- but the state does not exist to make statements. 

Wars shouldn't happen simply to communicate that yes, we're taking this seriously. Arrests shouldn't happen simply to demonstrate a heightened level of suspicion. When states delegate the right to use violence on their behalf, those people to whom we've delegated have a responsibility to do what's effective, and what causes the least collateral damage, not what is most satisfying to those aggrieved by the state's failure to act.

I do not trust the laws which would empower the police to pursue white supremacists with omnipresent surveillance and indiscriminate violence. I do not trust the police implementing those laws to use them to the benefit of black Americans. I do not understand why anyone else would, other than -- perhaps -- as a metaphorical howl of despair that the official outlets of state violence can't be trusted to deploy that violence reasonably. 
57 comments on original post
21
Add a comment...

Colin McMillen

Shared publicly  - 
 
 
Well, folks. I'm finally back from my enforced blogging hiatus with a bit of a doozy. There were quite a number of trailers coming out of E3 this year featuring playable female characters, and while that definitely represents a shift in the right direction I also wanted to take a look at why these games are more of a mixed bag that it would seem at first glance.
Another E3 has come and gone! And it seems that, after the debacle of last year's E3, that developers are making a bit more of an effort to not fail at easily-preventable sexism. While still nowher...
View original post
3
Add a comment...

Colin McMillen

Shared publicly  - 
 
Apparently, being a mostly-straight, childless, dude still doesn't keep me from crying in public when the next generation displays signs of being empirically more awesome than their predecessors.
59
1
Chris Shabsin's profile photoGreg S's profile photoRiff Zifnab's profile photoTrevor Schroeder's profile photo
11 comments
 
Colin: I'm sorry I crapped up your lovely post. \:

Greg S: If you want to continue this discussion feel free to start a new post and add me.
Add a comment...
In their circles
458 people
Have them in circles
10,229 people
blue blue's profile photo
Jeremy Zimmer's profile photo
Tài Liệu Thư viện's profile photo
José Albert Chachapoyas Sánchez's profile photo
Shannon Poole's profile photo
Awet Ayu's profile photo
Murat Armağan's profile photo
Thu Tu Koelling's profile photo
Muzafar Ali's profile photo

Colin McMillen

Shared publicly  - 
 
 
"What Google’s doing, in these cases, is using its deep pockets in the interest of broader social ends, with seemingly little concern for short-term returns. This strategy has historical precedent. In the early years of the American republic, there was little appetite for government spending on public works, like roads and canals. But the country needed better roads to facilitate the growth of trade and commerce. So the states turned to private companies, which built turnpikes that they then operated as toll roads. In the late 18th and early 19th centuries, hundreds of these companies invested millions of dollars in laying thousands of miles of road, in effect providing the basic infrastructure for travel in the United States.

What’s interesting about these companies is that while they were, in theory, for-profit, and while they had shareholders, in most cases there was no expectation that they would actually turn a profit in operating the roads—tolls were kept low enough to encourage traffic and commerce. Instead, the shareholders—who were typically local merchants and manufacturers—saw their investments in turnpikes as a way to collectively provide a public good that, not incidentally, would also deliver benefits to them as business owners and consumers. They knew, of course, that other businesses would benefit from these roads even if they didn’t invest in them (the nature of a public good being that everyone can use it). But that didn’t mean the investment wasn’t worth making. It’s hard not to see a similar logic underlying much of what Google does"

h/t +Chris Jones​
3 comments on original post
36
6
Chung Wu's profile photoSteve S's profile photoMatt Crain's profile photoSam Romano's profile photo
3 comments
 
+Christopher Tate google is only one business. The whole industry needs to get into the lobby/(bribery) game like the gun industry, oil/gas/petro-chem, pharmaceutical, and defense contractors.

Google could lead by example in this case but their performance compared to other lobby groups is woeful.

...to claim those other PACs and lobby groups play by any code of ethics such as "don't be evil" is a pipe dream. You don't want to know how the sausage is made but there are necessary evils when it comes to politics.
Add a comment...

Colin McMillen

Shared publicly  - 
 
 
Despite this article's title, it's about more than just women of color in technology: it's about recruiting and retaining people from underrepresented groups across the board. And that's something extremely important to the success of any technology company.

Why? There are three major reasons.

(1) Diverse groups avoid stupid product mistakes. This is in literally every sense of the word "diverse:" if you have people from different groups in your team, they'll notice – and you'll prioritize – problems that you never would have spotted otherwise. If your system doesn't work for the deaf and someone on your team is deaf, or if it requires hitting tiny affordances all the time and you have someone with a motion disability, you're never going to ship it that way, and that means more users. If your system has a price structure, or a branding, or a visual style that would never appeal to users outside of Silicon Valley, you'll catch that if people on your team are from a very different world. If women experience a different kind of abuse on your system than men do, then you'll build entirely different protections into your system if there are women in the room when you're making the design decisions.

The key point is that these are just examples: nobody can predict what an extra set of eyeballs, especially different eyeballs, will catch. The one thing that's reliable is that each set of eyeballs – not just working grunt jobs, but in the core decision-making process – means you don't make a mistake that shuts out a bunch of potential customers.

(2) Diversity interrupts groupthink. It's really easy for a room full of similar people to start to talk in similar ways. Not only do you not make the right decisions, you don't even realize there are decisions that you're implicitly making. More different eyes prevent that.

(3) You get to hire the best people. People who haven't been in this game very long think "Recruit minorities? You mean lower the bar!" People who have played this for a while hear that and think "Sucker."

The thing about structural racism/sexism/etc. is that a lot of people from the various underrepresented groups don't have the "traditional signifiers" of being good. They won't have gone to the top-tier schools, or they won't have any contacts, or their job history will be so-so. What you quickly learn in engineering, though, is that these signifiers are simply signals that you use when trying to find good people – and overall, as signals, they kind of suck. Terribly.

I've lost count of how many people I've interviewed who came from top-tier schools and had a glowing résumé and couldn't think an independent thought or design a system on their own to save their lives. Top-tier schools don't provide a systematically better education in CS; often, CS departments are so mathematically inclined that students that don't actively go the extra mile come out with a degree in theory and no ability to code. They used to claim that they were "filtering out the best of the best," but in practice, they do a lot of that filtering starting from "people with enough contacts to get in." 

Job histories are sometimes useful, sometimes not, especially in an era where so many people end up unable to find a job for months or years at a stretch anyway. 

References are great, but they're only a positive signal: the lack of references tells you nothing.

And the important thing is, that unless you're a tiny company hiring a temp, or hiring a senior specialist, you shouldn't be hiring for experience: you should be hiring for brains. You can teach CS; you can't teach smart.

What this means is that among these "underrepresented groups," there are a bunch of smart people out there who, lacking these traditional signifiers, aren't getting the right job offers. And that means smart people that you can hire. Lots of them. All you have to do is hire them and treat them with respect.

(As a side note: I attended GHC, the biggest annual conference for women in CS last year, for recruiting purposes. The quality of people looking for jobs there was insane compared to any other CS event.)


But.... if you want to hire and retain these people, you have to make an active effort. This open letter has a bunch of specific suggestions in it which I personally think are all individually excellent: I endorse these ideas wholeheartedly.

(NB: It also makes several statements about how various companies do things. I have it on good authority that several of these statements are incorrect, but I have no personal knowledge either way and so am neither affirming nor negating that part. My endorsement of this letter is about all of the courses of action it favors, which I think are excellent ideas; on the rest, I have no opinion)

I will add: In my groups, people of all genders, races, and backgrounds are not only welcome but actively desired. This is the case now and will continue to be the case in every team I run in the future.

Thanks to +Erica Joy for pointing me at this great letter.

[DISCLAIMER: I am writing this post in my personal capacity and am not speaking on behalf of Google. I make no assertions as to the truth or falsity of any of the claims of fact made within the letter, nor of any conclusions of law. Those of you who have been in the field for a while know why I have to state this, too]
81 comments on original post
12
1
Bridget Spitznagel's profile photoJustine Alexandra Roberts Tunney's profile photoDaniel Egnor's profile photoEddie Kay's profile photo
4 comments
 
Call-out mental illness stigma: When someone admits to an anxiety disorder, it doesn't mean you have to distrust their fact checking or whatever else +Justine Alexandra Roberts Tunney​​ might be implying. It just means they've been honest about a difficulty most people would hide.

In any case, it seems like the article should be able to stand on its own merits.
Add a comment...

Colin McMillen

Shared publicly  - 
 
Percents and fractions are hard.
 
Use .33333 percent of your leap second this year to learn what a leap second is. Super meta.
95 comments on original post
22
1
Jesse Morris's profile photoAdam Liss's profile photoDavid Saff's profile photoEvgeni Belin's profile photo
4 comments
 
Precision is apparently also hard?
Add a comment...

Colin McMillen

Shared publicly  - 
 
1) This makes it crystal-clear that the 4 "liberal" justices always vote as a bloc, and therefore need exactly one "defector" from the other 5 to agree with them. Usually that is Kennedy, but twice it was Roberts, and once it was Thomas.

2) The Roberts-Alito-Scalia bloc is less consistent, but in 12 of the 15 cases, all three voted the same way.

3) I was surprised how often Thomas votes not-with-his-supposed-bloc — in 5/15 of the votes, he voted contrary to the majority opinion among {Roberts, Alito, Scalia}; and in the "Confederate Flag and Free Speech" case, his vote was the decisive one.
 
This website has a great summary of what the cases were, and who voted how. I found it really helpful in trying to figure out what was going on.

http://www.nytimes.com/interactive/2015/us/major-supreme-court-cases-in-2015.html
2 comments on original post
10
2
Steve S's profile photoChristopher Tate's profile photoEddie Kay's profile photoChung Wu's profile photo
6 comments
 
Some years ago the joke was that Clarence Thomas was in the conservative bloc unless porn was involved, in which case he was all for it.
Add a comment...

Colin McMillen

Shared publicly  - 
 
More photos from Cambridge city hall.
11
Add a comment...

Colin McMillen

Shared publicly  - 
 
You don't have to be a Supreme Court justice. Noone is forcing you to.
50
2
Steven Johnson's profile photoEd Holden's profile photoIsaac Clerencia's profile photoJeff Grimes's profile photo
17 comments
 
Totally agree with +Jason Riek​​. This article articulates it well:

http://www.vox.com/2015/6/26/8851495/same-sex-marriage-scalia

While Scalia's warnings are eloquent, his actions routinely prove his point rather than heed it. On Thursday, for instance, Scalia's furious dissent lamented his inability to rip health insurance subsidies away from 6.4 million people because he had decided to forget how to read words in context.

The Supreme Court is one of the scariest entities in Washington; it's easy to forget on occasions when they just happen to be not-assholes by at least 5-4.
Add a comment...
Colin's Collections
People
In their circles
458 people
Have them in circles
10,229 people
blue blue's profile photo
Jeremy Zimmer's profile photo
Tài Liệu Thư viện's profile photo
José Albert Chachapoyas Sánchez's profile photo
Shannon Poole's profile photo
Awet Ayu's profile photo
Murat Armağan's profile photo
Thu Tu Koelling's profile photo
Muzafar Ali's profile photo
Work
Occupation
Software Engineer
Employment
  • Google
    Staff Software Engineer, 2009 - present
  • reCAPTCHA
    Co-Founder, 2007 - 2009
Places
Map of the places this user has livedMap of the places this user has livedMap of the places this user has lived
Currently
Boston
Previously
Minneapolis - Pittsburgh - Tokyo
Story
Bragging rights
I proposed to my wife using obfuscated Perl code. My Erdős number is 4.
Education
  • Carnegie Mellon University
    Ph.D. Computer Science, 2003 - 2009
  • University of Minnesota
    B.S. Computer Science, 1999 - 2003
  • Robbinsdale Cooper High School
    1995 - 1999
Links
Contributor to