Planet (Former) Advogato

This is a complement to Advogato, it is an aggregation of blogs of those who used to post on Advogato, but for one reason or another moved their blog from Advogato. It is provided as a service to those who would like to read the "greater Advogato" community.

This site works only as a Planet, it aggregates the post only, to comment on a blog entry, click on the title or time to go to the blog entry on the original site, hopefully it will have a comment facility.

July 26, 2014

Don Marti [dmarti]

Newspaper dollars, Facebook dimes

(updated 27 Jul 2014: add Gannett ad revenue)

Hard to miss the Facebook earnings news this week.

Facebook earnings beat expectations as ad revenues soar

Facebook Beats In Q2 With $2.91 Billion In Revenue, 62% Of Ad Revenue From Mobile, 1.32B Users

Let's take a look at those numbers. (I'd like to fill in more and better data here, so any extra sources welcome.)

Mobile ads: 62% of ad revenues.

Total US ad revenue: $1.3 billion.

Which would make mobile US revenue about 800 million. (Other countries are heavier on mobile, so this might even be high.)

Americans spend 162 minutes on a mobile device per day of which 17% is Facebook. So figure about 28 minutes per day on average. (Average of all US "consumers", not just mobile or Facebook users.)

That's double the time spent reading the printed newspaper.

US users spend an average of 14 minutes/day on printed newspapers. (Average of newspaper readers and non-readers. Just print, not web or mobile.)

But how are newspapers doing with the ad revenue?

Even after a sharp decline, newspaper print ad revenue in the USA is at $17.3 billion/year. That's the 2013 number, so it's reasonable to expect it to continue to come down as newspaper-reading time continues to decline.

Let's say it comes down another 10 percent for this year (which is faster than trend, and Gannett's print advertising is only down 6% this quarter compared to a year ago) and take a quarter of that. That's $3.9 billion.

So the newspaper brings in more than four times as much ad money by being in front of users for half the time. The newspaper completely lacks all the advanced behavioral targeting stuff, and Facebook is full of it.

What's going on here? Why is Facebook—the most finely targeted ad medium ever built—an order of magnitude less valuable to advertisers than the second-oldest low-tech ad medium is?

Here's my best explanation so far for the "print dollars to digital dimes" problem.

Advertising is based on a two-way exchage of information. You, the reader, give advertising your attention. Advertising gives you some information about the advertiser's intentions. That's often not found in the content of the ad. The fact that it's running in a public place at all is what builds up your mental model of the product, or brand equity.

On the other hand, advertising that's targeted to you is like a cold call or an email spam. You might respond to it at the time, but it doesn't carry information about the advertiser's intentions. (For example, you might be the one sucker who they're trying to stick with the last obsolete unit in the warehouse, before an incompatible change.)

As Bob Hoffman, Ad Contrarian, wrote, Online advertising has thus far proven to be a lousy brand-building medium. Walk through your local supermarket or Target or Walmart and see if you can find any brands built by online advertising. So what is web advertising good for? Thus far, it has been effective at search and moderately effective at a certain type of direct response.

Without the signaling/brand building effect, those targeted Facebook ads don't pull their weight, and come in at less valuable than newspaper ads.

I'm not saying we should go back to dead trees, but clearly mobile is leaving money on the table here. What's the solution? Paradoxically, it's going to have to involve some privacy tech on the user's end—preventing some of the one-sided data flow towards advertisers in order to introduce signaling effect.

More: Targeted Advertising Considered Harmful

July 26, 2014 02:43 PM

July 24, 2014

Thomas Thurman [marnanel]

Gentle Readers: catch them, Rimeq

Gentle Readers
a newsletter made for sharing
volume 1, number 15
24th July 2014: catch them, Rimeq
What I’ve been up to

I read a choose-your-own-adventure science fiction book when I was little. It concerned the efforts of an alien named Rimeq to take over the world, and the hero's efforts to stop him. This was made more difficult because Rimeq possessed the ability to move objects around with his mind (telekinesis). The only part which has stayed in my head is towards the end, when the hero has reached Rimeq's room but Rimeq has paralysed him by telekinesis, the police have been stopped similarly, and so have the spaceships bringing help, and the stress is showing on Rimeq's face. Finally the hero manages to take some rings off his fingers and throw them at Rimeq, shouting, "Catch them, Rimeq, they're grenades!" This is the final straw; the stress on Rimeq's mind is too much, and he is taken away catatonic.

So as I mentioned earlier, we have been moving house, and several moments have made me think, "Catch them, Rimeq"-- in particular, I meant to put out an edition of Gentle Readers on Monday as usual, but exhaustion won. Sorry for the interruption in service; meanwhile, I've been very encouraged by the messages I've had telling me how much you enjoy reading Gentle Readers.

Many people are due public thanks for helping us get through the last week. In particular, I want to thank the people of St John's church, Egham; as the obstacles to getting moved grew more and more formidable, so more and more people from St John's turned up unasked to help. We couldn't have managed without you. Thanks also go to the Gentle Reader who offered a garage when the movers needed to deliver before the landlord could give us the key. And thanks to the people from the Runnymede Besom, who turned up to take away some furniture we'd donated, but then came back later to help clean up. That's what love in action looks like, and I'll do my best to pay it forward. Thank you all.

A poem of mine

THE ECHOES OF AN AMBER GOD
(T54)

Electric sparkles in your touch,
the echoes of an amber god.
You fill my batteries with such
electric sparkles in your touch,
that Tesla would have charged too much
and Franklin dropped his lightning-rod:
electric sparkles in your touch,
the echoes of an amber god.

A picture

I was going to draw you a cartoon as usual, but my tablet is still packed away. Instead, here are some photos I took when I was working in London earlier this year.

http://thomasthurman.org/pics/clapham-junction
Trains in the sidings at Clapham Junction, the busiest railway station in Britain.
More than a hundred trains an hour come through.

http://thomasthurman.org/pics/binder
The tombstone of Jason Binder:
"He respected all living things. His inspiration lives on."
And it lives on with me, too, even though his epitaph is all I know about him.

 

Something from someone else

Does this one really need an introduction? Well, if you've never seen it before, then you have the joy of seeing it for the first time; the Guardian has a decent analysis if you're interested in digging into it. "Baggonets" is an archaic form of the word "bayonets", and Kensal Green is a large London cemetery, one of the magnificent seven. There is a pub called "Paradise" near there now; it was named for the poem.

THE ROLLING ENGLISH ROAD
by G K Chesterton

Before the Roman came to Rye or out to Severn strode,
The rolling English drunkard made the rolling English road.
A reeling road, a rolling road, that rambles round the shire,
And after him the parson ran, the sexton and the squire;
A merry road, a mazy road, and such as we did tread
The night we went to Birmingham by way of Beachy Head.

I knew no harm of Bonaparte and plenty of the Squire,
And for to fight the Frenchman I did not much desire;
But I did bash their baggonets because they came arrayed
To straighten out the crooked road an English drunkard made,
Where you and I went down the lane with ale-mugs in our hands,
The night we went to Glastonbury by way of Goodwin Sands.

His sins they were forgiven him; or why do flowers run
Behind him; and the hedges all strengthening in the sun?
The wild thing went from left to right and knew not which was which,
But the wild rose was above him when they found him in the ditch.
God pardon us, nor harden us; we did not see so clear
The night we went to Bannockburn by way of Brighton Pier.

My friends, we will not go again or ape an ancient rage,
Or stretch the folly of our youth to be the shame of age,
But walk with clearer eyes and ears this path that wandereth,
And see undrugged in evening light the decent inn of death;
For there is good news yet to hear and fine things to be seen,
Before we go to Paradise by way of Kensal Green.

Colophon

Gentle Readers is published on Mondays and Thursdays, and I want you to share it. The archives are at http://thomasthurman.org/gentle/ , and so is a form to get on the mailing list. If you have anything to say or reply, or you want to be added or removed from the mailing list, I’m at thomas@thurman.org.uk and I’d love to hear from you. The newsletter is reader-supported; please pledge something if you can afford to, and please don't if you can't. Love and peace to you all.

This entry was originally posted at http://marnanel.dreamwidth.org/307056.html. Please comment there using OpenID.

July 24, 2014 11:56 PM

July 23, 2014

Bastien Nocera [hadess]

Watch out for DRI3 regressions

DRI3 has plenty of necessary fixes for X.org and Wayland, but it's still young in its integration. It's been integrated in the upcoming Fedora 21, and recently in Arch as well.

If WebKitGTK+ applications hang or become unusably slow when an HTML5 video is supposed to be, you might be hitting this bug.

If Totem crashes on startup, it's likely this problem, reported against cogl for now.

Feel free to add a comment if you see other bugs related to DRI3, or have more information about those.

Update: Wayland is already perfect, and doesn't use DRI3. The "DRI2" structures in Mesa are just that, structures. With Wayland, the DRI2 protocol isn't actually used.

July 23, 2014 12:18 PM

Eric Anholt [anholt]

vc4 driver month 1

I've just pushed the vc4-sim-validate branch to my Mesa tree. It's the culmination of the last week's worth pondering and false starts since I got my first texture sampling in simulation last Wednesday.

Handling texturing on vc4 safely is a pain. The pointer to texture contents doesn't appear in the normal command stream, and instead it's in the uniform stream. Which uniform happens to contain the pointer depends on how many uniforms have been loaded by the time you get to the QPU_W_TMU[01]_[STRB] writes. Since there's no iommu, I can't trust userspace to tell me where the uniform is, otherwise I'd be allowing them to just lie and put in physical addresses and read arbitrary system memory.

This meant I had to write a shader parser for the kernel, have that spit out a collection of references to texture samples, switch the uniform data from living in BOs in the user -> kernel ABI and instead be passed in as normal system memory that gets copied to the temporary exec bo, and then do relocations on that.

Instead of trying to write this in the kernel, with a ~10 minute turnaround time per test run, I copied my kernel code into Mesa with a little bit of wrapper code to give a kernel-like API environment, and did my development on that. When I'm looking at possibly 100s of iterations to get all the validation code working, it was well worth the day spent to build that infrastructure so that I could get my testing turnaround time down to about 15 sec.

I haven't done actual validation to make sure that the texture samples don't access outside of the bounds of the texture yet (though I at least have the infrastructure necessary now), just like I haven't done that validation for so many other pointers (vertex fetch, tile load/stores, etc.). I also need to copy the code back out to the kernel driver, and it really deserves some cleanups to add sanity to the many different addresses involved (unvalidated vaddr, validated vaddr, and validated paddr of the data for each of render, bin, shader recs, uniforms). But hopefully once I do that, I can soon start bringing up glamor on the Pi (though I've got some major issue with tile allocation BO memory management before anything's stable on the Pi).

July 23, 2014 12:41 AM

July 22, 2014

Don Marti [dmarti]

How to beat adtech fraud: REGISTER ALL HUMANS

Ted McConnell, on AdExchanger: Advertising Fraud: It’s Time For Asymmetrical Warfare.

When you have an enemy that’s shape-shifting, agile, belligerent, invisible, greedy, fast and brilliant, you have a problem. Welcome to what military strategy people call asymmetrical warfare. It looks like terrorism. They lie about their identity. They only have to be right once. There are no lines in the sand. You can’t tell them from the good guys. They adapt.

I's actually worse than that. The best fraud rings only have to be better than the worst ad networks. The fraud perpetrators get to pick which network to attack, while the network doesn't get to pick which fraud perpetrators it deals with. The feedback for fraud is relatively quick. It's cheap and easy to try it on a small scale by buying or generating a little bit of bad traffic and seeing what happens. It's easy to decouple the parts of fraud that you're good at from the parts that you need help on, because that's how adtech is networked to begin with. Finally, the expected consequences of failure are small.

Where this piece gets problematic is in suggested solutions for dealing with the adtech fraud problem while keeping the adtech system intact. (Adtech, privacy, and fraud control, you can only have two.) Of course, this means abandoning privacy.

For example, "Make a publicly provided 'white list' of humans, accessible as a service to all transactions," and "tighten up Internet access...make sure an antivirus is in place." So in order to beat adtech fraud, McConnell wants to have (1) a white list of all humans and (2) control over all client systems (to verify that antivirus). Even the DRM maximalists didn't get that much.

And what happens while this perfect system of total control is being rolled out? Older clients, and humans who aren't on the white list of humans, will still be out there, so most of the fraud gets to continue. And by the time the system of control is in place, someone will subvert it for legit reasons.

If total Internet lockdown isn't going to happen, how do you beat fraud? A better answer is to turn the privacy up, not down: Adtech fraud: you can't cheat an honest man.

Bonus links:

Jon Udell: It’s time to engineer some filter failure

Atul: Does Privacy Matter?

Tim Peterson: Angry Birds Maker Rovio Points Finger at Ad Networks Over NSA Data Leak

Randall Rothenberg, president and CEO of the Interactive Advertising Bureau: IAB Head: 'The Digital Advertising Industry Must Stop Having Unprotected Sex' (via The Drift from Upstream)

David Rogers: Bad adbots and the vanishing CMO

Robin Hanson: Why Do Firms Buy Ads?

Ted Dhanik: We're All Responsible for Click Fraud and Here's How to Stop It

Doug Weaver: Dead internet ideas: The "right" to target

July 22, 2014 12:20 PM

Mary Gardiner [hypatia]

Opt-in Creative Commons licencing plugin for WordPress?

Does anyone have a recommendation for an opt-in Creative Commons licencing plugin for WordPress. That is, one where the default state is not to CC licence something, but when some action is taken, an individual post or page can be so licenced.

As background: I have no desire to write, maintain, or even debug a WordPress plugin. I want to know if there is something for this use case that Just Works.

I want opt-in, because it is too hard to remember, or to train others, to find an opt-out box when posting, and thus end up CC licensing things that weren’t intended to be, or can’t be, released under such a licence.

Some options I’ve already looked into:

WP License reloaded: was pretty much exactly what I wanted but doesn’t seem to be actively maintained and is now failing (possibly because the site in question is now hosted on SSL, I’m not sure, see above about not being interested in debugging).

Creative Commons Configurator: seems to be the most actively maintained CC plugin, but seems to be opt-out, and even that was only introduced recently.

Creative Commons Generator: opt-out.

Easy CC License: perhaps what I want, although I’d rather do this with an options dialogue of some kind than a shortcode.

July 22, 2014 07:11 AM

July 21, 2014

Thomas Thurman [marnanel]

enter price

I am in the chemist's waiting for a prescription to be filled, and eavesdropping.

Customer, to assistant: How much is this?
Assistant: (scans it repeatedly) Dunno.
Pharmacist: What's up?
Assistant: Every time I scan this, it just says "enter price", "enter price".
Marn: (under breath) These are the voyages of the Starship Enter Price...
(Pharmacist laughs. Assistant looks confused.)
Pharmacist: Well, *I* thought it was funny.

This entry was originally posted at http://marnanel.dreamwidth.org/306752.html. Please comment there using OpenID.

July 21, 2014 08:57 PM

July 20, 2014

Mary Gardiner [hypatia]

The Sydney Project: Luna Park

This year is my son’s last year before he begins full time schooling in 2015. Welcome to our year of child-focussed activities in Sydney.

Luna Park entrance

by Jan Smith, CC BY

Luna Park is, honestly, essentially cheating on this project. Do children like amusement parks? Yes. They do. There you go.

In addition, I think four years old is basically about the right age for them. It’s old enough that children are aware that a giant painted face, tinkly music, and carousels aren’t a completely normal day in the world, young enough that the carousel is still just as magical as the dodgem cars. And too young to have horror-film associations with amusement parks, I think that helps too.

Luna Park ferris wheel

by Kevin Gibbons, CC BY

It’s also more accessible to a four year old than some more thrill-oriented parks. V isn’t scared of heights or speed, so he loves the Coney Island slides, and was annoyed to find out that he was too short for the Ranger (the ship you sit in that gets spun upside down about ten stories in the air) and the free-fall ride. He is, however, apparently afraid of centrifugal force parallel to the ground, and refused to go on any “octopus” rides.

Even the four year old who wants to go on the free-fall ride is still young enough for, well, frankly dinky rides like the train that goes around about five times in a circle while you pretend to drive it, and the space shuttles that turn in gentle circles and which slowly go up and down when you press a button. His big draw is the ferris wheel, which I found fairly horrifying this time as I read the signs about keeping limbs inside to him and then had to answer a lot of questions about “why? why do I have to keep my limbs inside?” while giant pieces of metal calmly whirled past us with their comparatively infinite strength. In a similar vein, V also enjoys the roller coaster past all reason and sense, whereas Andrew and I react with “this seems… flimsy…” (I love coasters, but I like them to look overengineered).

Luna Park, where there's still a space shuttle

The only things V really didn’t like were the organised dancing groups who were encouraging children to learn their (cute!) 1930s-ish moves, and the process of choosing a child from a hat to press the lever to light up the park at night (he refused to let his name be entered), because there’s some specific types of performative attention that he really loathes. But there’s plenty of children gagging to dance along and to light up the park that an objector goes unnoticed. It’s not coercive fun.

Cost: entry is free. Rides aren’t, an unlimited rides pass for the day starts at $29.95 for a young child and goes to $49.95 for a tall child or an adult. There are discounts for buying online. (The entry is free thing sounds really useless, but it’s actually good if you have several adults, not all of whom are interested in the rides and/or are looking after babies.)

Recommended: indeed. We’ve considered getting an annual pass, in fact.

More information: Luna Park Sydney website.

Disclosure: because of a prior complaint to Luna Park about opening hours (we showed up several months ago at 2:15pm to find that an advertised 4pm closure had been moved to 3pm), we were admitted free this time. No reviews were requested or promised in return for our admission.

July 20, 2014 11:52 PM

Hubert Figuière [hub]

Going to Guadec

For the first time since 2008, when it was in Istanbul, I'm coming to Guadec. This time it is in Strasbourg, France. Thanks to a work week scheduled just before in Paris.

I won't present anything this year, but I hope to be able to catch up a bit more with the Gnome community. I was already at the summit last fall, as it was being held in Montréal, but Guadec participation is usually broader and wider.

July 20, 2014 10:58 PM

July 19, 2014

Thomas Thurman [marnanel]

kids' fascination with death

As a littl'un, my daughter was interested not only in Ancient Egypt but also in the Soap Lady in the Mütter museum-- a corpse which has become entirely saponified, turned to the soapy substance called adipocere. One day, when my daughter was about five, I was sitting reading while she was playing in the park, and eavesdropping on her conversation with another girl:

Other Girl: "Do you know what happens to you when you die?"
Rio: "Yes. You turn into soap."
Other Girl: "No... you turn into stone. I know because my grandma died and I touched her and she was as cold as a stone."

This entry was originally posted at http://marnanel.dreamwidth.org/306537.html. Please comment there using OpenID.

July 19, 2014 08:06 PM