Forgot your password?

Comment: Re:Of course not! (Score 1) 64

by swillden (#48446263) Attached to: 2014 Hour of Code: Do Ends Justify Disney Product Placement Means?

It's reprehensible that they leverage this incredibly popular brand to teach girls to code when they could be using it to sell Happy Meals and next year's landfill fodder. Shame, shame!

You're missing the point: Disney is exploiting the incredible popularity of Hour of Code among young girls in order to boost their poorly-performing movie.

That might be shameful if it weren't so completely ludicrous.

Comment: Re:Invite link? (Score 1) 284

by swillden (#48438233) Attached to: Google Launches Service To Replace Web Ads With Subscriptions

After all, they have like 98% marketshare, while the 2% belong to those more questionable networks (the ones that advertise for sites that Google won't touch - e.g., torrent sites and the like).

Actually, 33%. They're by far the biggest single player, but aren't anywhere close to 98%. Google's share of mobile ads is larger, at 56%. (that's 2013, but things haven't changed much in 2014, and I couldn't find a 2014 link that included both all digital and mobile ads).

Comment: Re:Here we go again (Score 1) 482

by gmhowell (#48430157) Attached to: As Amazon Grows In Seattle, Pay Equity For Women Declines

Do you know anything about running a business? In a service industry, people costs are often a huge portion of a company's overall costs. Minor changes in that structure can have major impact on the bottom line.

Even in traditional manufacturing jobs, where a large percent of the costs are tied up in capital and materials, a modest change in employee costs filters through. Just ask GM and Chrysler.

Comment: Re:Well that's a start... (Score 2) 158

by swillden (#48423097) Attached to: Number of Coders In Congress To Triple (From One To Three)

<counter-pedantic>Not in C++.</counter-pedantic>

Eh? The C++ standard explicitly forbids "void main()". From the standard:

An implementation shall not predefine the main function. This function shall not be overloaded. It shall have a return type of type int, but otherwise its type is implementation-defined. All implementations shall allow both of the following definitions of main:

int main() { /* ... */ }


int main(int argc, char* argv[]) { /* ... */ }

Comment: Re:Migration away from Google? (Score 1) 382

WHY IT ISN'T THE DEFAULT - is anyone's guess.

It's quite obvious, actually... it's not the default because it doesn't work as well for most people. Verbatim is good when you're searching for fairly specific terms, spelled correctly. If you're asking a more general question, with words that may appear in many variations, or if you don't spell well or are lazy, then the "new" Google works dramatically better.

I think a lot of complaints about Google search today, especially by people who have been around for a while, really boil down to the fact that the old search tricks don't work very well any more. In the early days of search we all learned how to create effective search queries, by picking carefully targeted search terms, combining them in particular ways, omitting any extraneous or "filler" words and lots more that make search queries look very different from natural language. But the search engines (or at least Google) have been changing along with the user base, which is now comprised of almost entirely non-technical people who haven't been using the web for long enough or heavily enough that they learned to compose searches that catered to the engines' weaknesses.

So, today, Google focuses on optimizing for the now-common case of search queries which are most often natural language questions, typed quickly and carelessly. The search engine tries hard to figure out what the user meant, rather than what they said. To those accustomed to being very precise and saying exactly what they mean, this is somewhat infuriating, because they don't want the machine to guess at what they meant, they told it what they meant. For the average user, though, who is more accustomed to dealing with people, who are good at guessing what is meant, the new system works much better.

Personally, I've adapted to the new reality. I tend to type complete sentences for my search queries, framed as questions, including typing the question mark (not because I think it's useful but just because I'm thinking a question sentence, so my fingers emit a question mark). I also don't worry much about typos. I find it works very well, often much better than what I can get with an "old-style" query, with or without "verbatim".

(Disclaimer: I work for Google, but on Android, not search. All of the above is just my personal experience plus speculation, not inside information.)

Comment: Re:No trust (Score 1) 535

by swillden (#48418597) Attached to: Debian Votes Against Mandating Non-systemd Compatibility

With the failure of this GR, it is clear that I can not trust Debian to ensure that systemd remains optional.

Why is this important to you? Serious question. I don't really have an opinion on it, myself, but it seems to me that all of the arguments against systemd are based on factual errors (e.g., that it's monolithic, and therefore not UNIXy) and inertia, or on defects that are clearly just packaging/configuration bugs. I found Russ Allberry's analysis pretty compelling. Why do you disagree?

I'm really wondering what I'm missing here, because this seems like much ado about nothing, and I haven't been able to get anyone who is really concerned about it to explain why it's really a big problem.

Comment: Re:its all about choice. (Score 1) 535

by swillden (#48418097) Attached to: Debian Votes Against Mandating Non-systemd Compatibility

Your comment confuses me.

You start by saying that the proposal, that packagers be required to maintain support for systems without systemd, is untenable. Then you point out that Debian should realize that users can code rc-init support for packages if they want to. I agree with all of that: Debian is going systemd, and shouldn't burden package maintainers with supporting non-systemd initialization, and users who don't like that can code rc-init scripts for the packages.

But then you say that Debian should give users the choice. Did you just finish pointing out the users do have the choice, since they can code it themselves if they want, and that the burden for this shouldn't be placed on maintainers?

Also, I think you meant to say "wealth", rather than "dearth" (which means a lack, not an abundance). But maybe you did mean dearth and I'm just not understanding what you're trying to say.

Comment: Re:should be banned or regulated (Score 1) 237

by swillden (#48407929) Attached to: Will Lyft and Uber's Shared-Ride Service Hurt Public Transit?

The reason we require insurance coverage for cabs is that we had many accidents in which people were severely injured, including pedestrians who never contracted with the cab driver, and it turned out that the cab driver didn't have enough insurance to cover them.

Which is why Uber now provides a $1M policy covering all of their drivers. Does that address that issue?

The reason we require a hack license is that, among other things, we want cab drivers to go through a police check to make sure they haven't committed crimes in the past.

Okay, but is there any evidence that actually accomplishes anything? Assuming that there is, and that it's useful, then why not just require a background check?

Uber claims they screen their drivers but it's up to them to convince us that they screen them as well as the hack bureau does.

Is there any evidence their screening is inadequate?

And what about a medallion? Bonding? And is race discrimination a problem at Uber or Lyft (or in any cab company these days)?

I do have to give you that you're the first to even attempt to dig into the underlying issues, though. Kudos for that.

Your own mileage may vary.