Death to Bullshit • Hacker News Board

Re: Death to Bullshit

#201

Post by superuser2 » Mon, Sep 07, 2015, 10:11 PM UTC

beagle3 wrote: post #197

superuser2 wrote: post #156

The vast majority of working adults in the US are employed in the making and selling of things people don't need. A world without bullshit is total, utter economic collapse. It's the end of capitalism. It's hundreds of millions of people with nothing to do all day and no way to sustain themselves, an inevitable civil war with the landlord class, and a revolution that manages to install a government's that quite possi…

> The vast majority of working adults in the US are employed in the making and selling of things people don't need. A world without bullshit is total, utter economic collapse. There's little difference between "things that people don't need", and "things that people may need but are done inefficiently". In the late 19th century, you might have said that "a world with cars means utter economic collapse of the whip-and…

>but that just means people (and jobs) are reallocated to more useful stuff.

No one argues that bullshit is constantly being iterated on. But cars are bullshit - i.e. we lived for a long time without them, and they needed to be advertised to take hold. Now simply having a car is entrenched bullshit, and the cutting edge is convincing people to buy expensive ones.

You don't need a car, but cars are necessary to facilitate the American Dream, also sold to us by advertising in the 40s/50s, of living in suburbs. Home ownership and the things that go with it are equally bullshit.

One of the first documented shifts in public opinion orchestrated by a PR firm is America's belief that roads are for cars instead of people.

Re: Death to Bullshit

#202

Post by shostack » Tue, Sep 08, 2015, 12:12 AM UTC

TeMPOraL wrote: post #82

fredkbloggs wrote: post #61

Earlier quoted context omitted.

> As much as I dislike some of the current trends (and I do digital media for a living mind you), I do have to point out that this stuff wouldn't be done if it didn't work. Are we sure of that? It might depend on our definition of "work". A lot of the people responsible for doing this stuff may just be doing it because if they admit that none of it works, they'll have nothing else to offer and will be out of a job. I…

> It's not terribly hard to make up data showing some kind of effectiveness, or just to misinterpret real data in a way that can convince others to let you keep doing it. For every marketing or advertising craftsman out there who builds something that works, gathers useful data, and interprets that data honestly, there are probably 20 people who have no clue and just keep churning out the same stuff because they don'…

Sounds like a low-quality shop or one that isn't being paid to do the strategic thinking around their measurement to determine "if it works."

As someone who is deep in the thick of that stuff, I'll be the first to admit the data is very fuzzy, and relies on a lot of statistics to really get a sense of whether things are working when there is not a strong direct response component (see: brand display advertising). Even the value things like paid search are entirely subjective depending on the attribution model you use and how you measure incremental lift.

But that isn't to say it is all voodoo that doesn't work. Its just that to measure it properly you need some very expensive technology and/or proper stats training.

Re: Death to Bullshit

#203

Post by TeMPOraL » Tue, Sep 08, 2015, 12:23 AM UTC

shostack wrote: post #202

TeMPOraL wrote: post #82

Earlier quoted context omitted.

> It's not terribly hard to make up data showing some kind of effectiveness, or just to misinterpret real data in a way that can convince others to let you keep doing it. For every marketing or advertising craftsman out there who builds something that works, gathers useful data, and interprets that data honestly, there are probably 20 people who have no clue and just keep churning out the same stuff because they don'…

Sounds like a low-quality shop or one that isn't being paid to do the strategic thinking around their measurement to determine "if it works." As someone who is deep in the thick of that stuff, I'll be the first to admit the data is very fuzzy, and relies on a lot of statistics to really get a sense of whether things are working when there is not a strong direct response component (see: brand display advertising). Eve…

> But that isn't to say it is all voodoo that doesn't work. Its just that to measure it properly you need some very expensive technology and/or proper stats training.

That's what I believe too. It does work when done properly, but most of the shops are not equipped to even begin to understand how to do it.

What kind of expensive technology do you mean in particular, if I may ask?

Re: Death to Bullshit

#204

Post by fredkbloggs » Tue, Sep 08, 2015, 2:03 AM UTC

shostack wrote: post #202

TeMPOraL wrote: post #82

Earlier quoted context omitted.

> It's not terribly hard to make up data showing some kind of effectiveness, or just to misinterpret real data in a way that can convince others to let you keep doing it. For every marketing or advertising craftsman out there who builds something that works, gathers useful data, and interprets that data honestly, there are probably 20 people who have no clue and just keep churning out the same stuff because they don'…

Sounds like a low-quality shop or one that isn't being paid to do the strategic thinking around their measurement to determine "if it works." As someone who is deep in the thick of that stuff, I'll be the first to admit the data is very fuzzy, and relies on a lot of statistics to really get a sense of whether things are working when there is not a strong direct response component (see: brand display advertising). Eve…

It might help to be clearer about what I'm getting at here. The question is not whether effective (and cost-effective) advertising exists, nor whether anyone anywhere knows how effective their advertising is. Clearly it does and some people do.

It's whether (a) buying ad impressions on a clickbait listicle (or other "bullshit" page) is cost-effective or effective at all, (b) whether writing and hosting such pages is profitable once all internal costs are properly considered, and (c) whether the advertising formats the author is complaining about are cost-optimal or even effective at all. I took the GP's claim that "it works" to mean that the answers to all three questions are an emphatic yes. I'm not so sure we know that, and I'm not so sure it's true. It seems likely that the answers might well be "almost never", "probably not", and "maybe sometimes", but factors other than profitability are at work that cause the people making the decisions to do these things even when an omniscient rational corporation would not. Improper use of statistics or poor data collection could certainly be an explanation for that, but I propose that it's more basic than that. Of course, I might also be wrong, and maybe all this bullshit does work, and works very well. That just doesn't seem like the slam dunk that was implied.

Re: Death to Bullshit

#205

Post by rjblackman » Tue, Sep 08, 2015, 4:40 AM UTC

icebraining wrote: post #191

rjblackman wrote: post #171

Earlier quoted context omitted.

you are mixing up communism and socialism. socialism and capitalism are not mutually exclusive. the scandinavian systems are called "nordic socialism" for a reason.

Socialism doesn't have an agreed-upon definition. For example, the World Socialist Movement will tell you that it entails an end to private property and to trades using money[1]. Whether the Nordic model is really socialist is a subject of intense debate among the Left (as usual). [1] http://www.worldsocialism.org/english/what-socialism

well the USSR under Gorbachev were considering moving to it (the nordic model) so I guess that gives some creedence to it's "left-ness"

Re: Death to Bullshit

#206

Post by shostack » Tue, Sep 08, 2015, 5:30 AM UTC

dang wrote: post #43

rayiner wrote: post #13

Earlier quoted context omitted.

The problem is that bullshit is a negative externality. It's cognitive pollution. Those who clutter the world with bullshit reap the rewards of nabbing the suckers who respond to it, but don't pay the costs of imposing those cognitive loads on everyone else. That leads to more bullshit than would be economically efficient. Take a simple example like billboards. If billboard advertisers had to pay every person whose l…

Thanks for putting into words what I've felt inchoately for years.

Out of curiosity, how do you balance the negative feeling you have towards that stuff against the companies you consider investing in? Do you simply choose not to invest in companies that promote these practices?

Re: Death to Bullshit

#207

Post by shostack » Tue, Sep 08, 2015, 5:39 AM UTC

TeMPOraL wrote: post #203

shostack wrote: post #202

Earlier quoted context omitted.

Sounds like a low-quality shop or one that isn't being paid to do the strategic thinking around their measurement to determine "if it works." As someone who is deep in the thick of that stuff, I'll be the first to admit the data is very fuzzy, and relies on a lot of statistics to really get a sense of whether things are working when there is not a strong direct response component (see: brand display advertising). Eve…

> But that isn't to say it is all voodoo that doesn't work. Its just that to measure it properly you need some very expensive technology and/or proper stats training. That's what I believe too. It does work when done properly, but most of the shops are not equipped to even begin to understand how to do it. What kind of expensive technology do you mean in particular, if I may ask?

This is going down the whole attribution rabbit hole, which is arguably the most challenge issue in the advertising/marketing space today.

Static attribution models like first touch, last touch, U-shaped, and even custom weighted models provided multiple lenses through which to view your data to see how it would be valued if more credit was given towards it (since historically things were viewed via last touch lenses for the most part). Google Analytics offers some great attribution tools for free if you're interested in checking those out.

Beyond that, static models have their weaknesses, so dynamic models such as the data-driven model in Google Analytics Premium are a nice step up. There are other solutions that handle this too such as VisualIQ, Adometry, Convertro, etc. Those were all recently acquired by Google, and others btw.

Then there is software that attempts to help you model incremental lift in addition to dynamic attribution models. I just demoed Adobe's econometric modeling tool in their Adobe Analytics package (that feature set might be in the higher-level version) and that was about the closest I've seen to date for doing it the way I consider the "right" way (ie. user-path level attribution and modeling for incremental lift using literally whatever external data you can feed into it).

I'm an ad tech nerd so let me know if you have any other questions.

Re: Death to Bullshit

#208

Post by shostack » Tue, Sep 08, 2015, 5:46 AM UTC

fredkbloggs wrote: post #204

shostack wrote: post #202

Earlier quoted context omitted.

Sounds like a low-quality shop or one that isn't being paid to do the strategic thinking around their measurement to determine "if it works." As someone who is deep in the thick of that stuff, I'll be the first to admit the data is very fuzzy, and relies on a lot of statistics to really get a sense of whether things are working when there is not a strong direct response component (see: brand display advertising). Eve…

It might help to be clearer about what I'm getting at here. The question is not whether effective (and cost-effective) advertising exists, nor whether anyone anywhere knows how effective their advertising is. Clearly it does and some people do. It's whether (a) buying ad impressions on a clickbait listicle (or other "bullshit" page) is cost-effective or effective at all, (b) whether writing and hosting such pages is…

Ok, let's get specific about listicle crap like this then (aka native advertising).

The formats, popularized by such ad networks as Taboola and Outbrain gained popularity because publishers needed ways to drive additional ad revenue without adding what appeared to be more display ads.

This happened to combine with the layout/design tactics of sites like Buzzfeed and HuffPo that have nailed the approach to keeping users engaged, clicking, and reading.

Coincidentally, this also helped provide a source of content for sites didn't really care about drawing the line between journalism and sales. Not that more respected publications haven't been trying them--just with tighter controls over the advertisers (perhaps via direct sales only).

Affiliates are all over these as expected, and if they can make it work, then that says something because affiliates operate on crap margins and have many other challenges.

That said, there's almost certainly a significant amount of dollars in there from companies just testing it because you need to constantly be testing new traffic sources and approaches. In fact, I always recommend companies carve out a test budget that they can just consider wasted, and if it happens to perform or they learn something, then it usually pays off in spades.

There are always those that are less savvy and tossing money away on wasted impressions. Then there are folks like myself who understand the limitations of the data, the nature of the medium, have their own audience data, etc. In many ways it parallels the differences between Wall St. and Main St. in the investing world.

Re: Death to Bullshit

#209

Post by walterbell » Tue, Sep 08, 2015, 5:59 AM UTC

shostack wrote: post #207

TeMPOraL wrote: post #203

Earlier quoted context omitted.

> But that isn't to say it is all voodoo that doesn't work. Its just that to measure it properly you need some very expensive technology and/or proper stats training. That's what I believe too. It does work when done properly, but most of the shops are not equipped to even begin to understand how to do it. What kind of expensive technology do you mean in particular, if I may ask?

This is going down the whole attribution rabbit hole, which is arguably the most challenge issue in the advertising/marketing space today. Static attribution models like first touch, last touch, U-shaped, and even custom weighted models provided multiple lenses through which to view your data to see how it would be valued if more credit was given towards it (since historically things were viewed via last touch lenses…

Do you know of any tools/studies/approaches which combine techniques from adtech attribution with techniques from online training, e.g. spaced repetition? E.g. are there timing models for saturation thresholds which cause users to ignore a message?

Re: Death to Bullshit

#210

Post by icebraining » Tue, Sep 08, 2015, 7:36 AM UTC

rjblackman wrote: post #205

icebraining wrote: post #191

Earlier quoted context omitted.

Socialism doesn't have an agreed-upon definition. For example, the World Socialist Movement will tell you that it entails an end to private property and to trades using money[1]. Whether the Nordic model is really socialist is a subject of intense debate among the Left (as usual). [1] http://www.worldsocialism.org/english/what-socialism

well the USSR under Gorbachev were considering moving to it (the nordic model) so I guess that gives some creedence to it's "left-ness"

Not even Lenin's USSR was considered Socialist by everyone (see the State Capitalism critiques, particularly by the anarchists), let alone the USSR under Perestroika!