Hacker News Reader

Comment by agentofoblivion | Hacker News Reader

Many of these commentors have it backwards. I know this because I’m actively working on a similar project at AWS. It isn’t that Amazon wants your data to help it make decisions, so it builds a clever product to get all your data. On the contrary, Amazon has so much damned data and years of experience turning that into a valuable asset, that it has started experimenting with ways to sell that value to other customers. The idea that it wants some mom and pops retail data is just silly. Our biggest problems are scaling the internal services we built 5 years ago to handle all the growth.

SageMaker was a ground floor service and you should expect to see many more services built atop it.

View on HN

People here mostly appear to be criticizing the quality of the algorithm, not the fact that Amazon wants the data.

so is this the equivalent of software as a service?

I mean, isn't that half of what AWS sells?

Amazon’s years of experience usually end up recommending you the same thing you just bought or just glanced at and never came back to the same category

This is very true. I'm forever being recommended things I searched for once for a parent or a friend that I have no other interest in.

If you were CEO of Amazon, what would you do? As much h as Amazon would love to have you but something you don't really need/want every day, it's not going to happen in substantial numbers. But they need something to keep brand awareness so you remember Amazon is the place to go for everything. Even complaining about the ads is still word of mouth for the store. Even of you don't like the recommendations, talking about Amazon is t bad PR for shoppers.

If you could solve the problem better, at scale, I’m sure they would pay you a lot of money to come solve it. Along with Netflix, Google, and all the other places that use standard algorithms designed to solve these sorts of problems.

This assumes internal politics are a meritocracy with appropriate levels of prototype funding.

I’m not sure that’s accurate — and there’s plenty of ways for organizational structure to strangle good ideas.

introduce temporal component into your prediction model, similar to this:

[0] https://en.wikipedia.org/wiki/Hierarchical_temporal_memory

I probably wouldn't pass the interviews at most of these places :-P

But if they pay me (well the consulting company I work for) enough I might fly in to hold some session along the lines of

- don't annoy power users for no good reason (a certain ad delivery company with an attached search engine might be interested in this)

- how to please all your customers at once - the forgotten art of making your software configurable. Bonus session: sensible defaults (audience: every modern software company)

- your customers might not be as smart as you but most of them aren't braindead - and other notes from the field (same audience)

- continue to be the best - by not nerfing your almost-perfect product (same audience)

- not storing data you don't need to serve your users [0] - a practical guide to not getting fined by EU while also getting rid of dumb disclaimers that won't help you anyway when regulators get fed up (audience: everyone who has a GDPR popup with default opt in to one or more likely more than 100 different trackers)

- why customers who have already bought a dishwasher won't buy a new one however much you advertise for them the rest of the month (and other great secrets from marketing 101). (Audience: most adtech companies and ad buyers)

- the long-term benefits of serving ads that actually provides value - or why you'd want to sell some ads to other companies than scammy dating sites even if they are currently the easiest ones to fleece by showing expensive ads to customers who aren't interested at all. (Ad-tech companies, and that big one with the attached search engine in particular.)

Master class:

- 1/3 Living in a global world: Europeans are also on the Internet and knows when movie and series releases are + other shocking truths.

- 2/3 increasing digital sales by putting the content up for sale - how to increase sales and customer happines all while reducing piracy just buy removing stupid geo limiting

- 3/3 the synergistic effects of making your content available to customers who wants to pay for it while it is still hot

[0]: Deliberately talking about users, not customers here to avoid the ever-helpful response of "if you are not paying you are not the customer, you are the product". And BTW, I agree, it just isn't always helpful.

I'm fairly sure that ignorance isn't the reason these companies aren't implementing your suggestions.

Well, I don't expect a sales boom but I struggle to come up with a better explanation for a couple of those.

Your issues with geo limiting around content are almost entirely an artifact of the publishing systems. Having had to work on those things, amazon does not actually want to waste time implementing them except that it's a requirement to get the content people want to buy. Put the blame where it belongs.

That part was about certain media/entertainment companies :-)

Then what does it have to do with recommendations?

It doesn't have to do just with recommandations but more generally all the big issues that the biggest companies have to deal with ;-)

Netflix actually did pay contest participants a lot of money ($1MM) to solve the problem better, at scale. But they didn't end up using the resulting algorithms.

https://www.wired.com/2012/04/netflix-prize-costs/

It seems odd though to suggest the same item I just bought over and over again.

Same as YouTube's algo now recommending already watched videos, over and over.

This is either being gamed by bots or someone at those cos is not dogfooding. And if all else fails, for the love of god, give me an option to train your dumb algo manually by blocking shit I don't want to see ever again.

> give me an option to train your dumb algo manually by blocking shit I don't want to see ever again.

Feel free to recommend my course posted below :-P

"how to please all your customers at once - the forgotten art of making your software configurable"

Sometimes it makes sense, for consumable goods or even non consumable things like storage containers, where you might buy more of the same item at different times. Other things like books I agree don't make sense.

Yeah YouTube's algo has become even more terrible, which I didn't think was possible.

Or another way to look at this is that if the problem can't be solved well - in a way that actually adds value - then maybe don't offer it at all.

I don't believe amazon's recommendation engine has ever delivered any value to me, and it would be really nice to just have all that crap removed.

If people stopped buying things they just bought, it would probably stop recommending them.

besides finding it odd how they select what to recommend me, including stuff I rated poorly that comes back up, the "See something new, every day" bar that displays shows item thumbnails that are not actually of items you will see if you click there.

so while they do somethings well there are just some odd choices made with how they present product. oh, with regards to that "see something new", reported it and got back that those thumbnails were not meant to represent actual products. call me confused

Reply on news.ycombinator.com