Hacker News Reader

Comment by throw_nbvc1234 | Hacker News Reader

Sure but do we need a push or pull based model. Ads are pushed into our attention. Google searches and "research" are pull based. If you solve the pull based model the king's doctor just looks it up, discovers something exists, and obtains the treatment.

Recommendations are similar to ads in this regard as well. Maybe blurring the lines between the two depending on how the algo is designed. I'd personally be in favor of recommendation type approaches with combinations of human curation, and an adjustable spectrum of push/pull based results; and the ability to swap between these on a whim as needed.

View on HN

Pull only works if you know what you're looking for. You'd end up researching race horses when you needed a moped.

Are you seriously saying that we need modern "push" marketing because doctors can't be trusted to google a cure for smallpox?

I agree that sometimes it's true that people need marketing to discover solutions, but, in the vast majority of cases I've encountered, marketing is self-serving, manipulative crap. For a concrete example, see cigarette advertising, corporations funding climate change denial, or most political advertising.

Not only that but it implies that specifically targeted forms of advertising (ie aanti privacy) are the ideal. Certainly an interesting perspective

They probably work for Google, because they did seriously say that and it is seriously ridiculous.

How does that adage go?

> It Is Difficult to Get a Man to Understand Something When His Salary Depends Upon His Not Understanding It

On the creator side, it is not so easy. If I think that someone is interested in my knowledge or product, I will write a book or produce it. But it only comes available until it lands in some shelf where it can be found. Hence, this is a push.

Alternatively, I could have waited until someone noticed my advertisement to start to write the book or produce the product. Pull.

So where is push and where is pull better?

I wasn't looking at this through the lens of creating content but rather starting from the point you "land in some shelf where it can be found". From this point it's about getting customers/audience which is where push/pull comes into play.

I see ads as purely a push based model; to the GP point, if it was worth looking for why wouldn't you already be looking for it via a pull based method. But you obviously don't know what you don't know. Either 100% push or 100% pull won't be optimal. This is where I see recommendations and being able to swap between a spectrum of XX% push and YY% pull algos is helpful.

For example physical music, i can walk into a record store and decide to buy a handful of records; the store owner can then see those records and recommend more that I could also enjoy and even ask me questions to make those more accurate. This isn't 100% push or pull and also has an element of human curation that inherently doesn't scale but is possibly the most effective approach. I can do the same thing online via discogs and get some automated recommendations. I have a choice between the two and depending if it's a genre I'm familiar with (human recommendation might give me something new) or a new one I'm trying to explore (most popular records recommended by discogs are a good starting point) i can choose one or the other.

I see solving this problem as a key one for the whole creator long-tail, 1k fans, etc... problem-space. Don't have the answers but those are some general ideas.

I'd argue that you definitely need a push-based model for preventative measures: vaccines, smoke detectors, Personal Protective Equipment to reduce the spread of disease.

If you don't know how serious a danger is, you won't start doing research on it until it actually occurs, at which point it is too late. It doesn't do much good to remind someone they need to have smoke detectors and check them regularly after the house burns down, or that a vaccine is available after they're infected.

Reply on news.ycombinator.com