Hacker News Reader

Comment by nkozyra | Hacker News Reader

They've been battling the widespread preference for chronological timeline for ... well, forever.

As the common belief goes, the "algorithmic" one is more profitable. I buy this, intuitively, because it seems natural that it'd be harder to sneak in something jarring if you were viewing a more traditional timeline.

But I can't fathom why seemingly no effort has gone into monetizing a chronological timeline. Why is it's foregone conclusion that it won't work with advertising the way the algorithmic was works?

View on HN

I don't think there's any issue monetizing a chronological timeline.

The issue for them I imagine is retention: with chronological, it's likely that when you visit Twitter the first few messages you see are not interesting and you leave. But if they put a tweet that's popular among your social group in front of you? Odds are that's interesting to you and you stick around a bit longer.

I dislike that, to be clear, but I imagine this is the human behavior for >90% of people.

Hotels can try boosting user retention, as measured by minutes spent in the room, by hiding guests' belongings while they are away.

That's pretty much the thinking behind layouts of malls and grocery stores.

Grocery stores put common items like milk way in the back, in order to have you traverse the whole store and get exposed to as many meters of shelf as possible. They generally also like to change placement of product, to make sure you don't learn the layout and get too efficient at navigating.

Malls try to balance enabling people to find what they are looking for, but also make sure they don't find it too easily so that people can wonder about and be exposed to as many stores as possible.

If you enter the Dubai Mall from the metro, there's a big sign that points to the left and says "Food Court". You'd get to the food court much faster if you turn right.

I wonder how much of this is an urban legend

There's a reason why refrigerated items are usually at the back: fridges usually face a wall. You can have a shelf at the back sometimes but that's not always possible

Also, for those quick purchases you can always go to somewhere closer, a gas station, etc.

In fact in Europe some places do have the quick purchases right in front of the store so you don't need to go around (Tesco as an example).

Nah, I saw this on my last shopping trip to a mega grocery. Bought basically all the items on my list on the right side of the store (produce, other food stuffs) and then had to walk past a half dozen aisles filled with random crap (non food) to get eggs from the dairy section.

Conversely at my local supermarket (major chain), the milk, meat, fruit and veg are all near the front of the store and easily accessible.

Also, why would your eggs be in the dairy section?

I think store documentation from the 'food store franchise' have been publicly shared in Sweden and had instructions on how you should place the milk.

The dairy section is the most infuriatingly obscure section in Swedish stores. To the point that I've seen it hidden away in a small subsection only accessible by a rather narrow entrance.

Don't forget IKEA.

I went the other day just to look at a dining table. It was only in the 3rd or 4th showroom. Of course, to get out I had to wander through the majority of like 13 or 14 other showrooms because you pretty much can only take the linear path to get to an exit.

Generally Ikea stores have a few shortcuts, but you usually have to pay attention to find them.

I heard from a friend that they're only there for IKEA staff to get around the store faster. I'm sure IKEA would otherwise do away with them.

> I heard from a friend that they're only there for IKEA staff to get around the store faster.

I doubt it. They are even marked on in store navigation. Example, https://miro.medium.com/max/1400/1*ZSJ1cE-FjFzNjIVz8l0a4Q.jp...

Besides the shortcuts mentioned by the sibling comment, you can also just go from the exit (yes, you can enter through the exit), straight to the warehouse.

You need to know the name of what you're looking for and then use their kiosks to find where the thing is in the warehouse and just pick it up directly.

This trick turns 1 hour trips through Ikea into 5 minute trips.

Didn't work for my usecase there. I specifically went to check out a table in person. Not necessarily buy it. It was only on display in a showroom, not the warehouse section.

Yeah, that sometimes happens. I don't really understand why they don't have all their products in the warehouse. Plus they don't deliver some stuff, despite the fact that they deliver almost everything. It's quite confusing.

Funny you mention the delivery thing. When I was in the store, I checked the website for a couple of tables. According to the site neither were deliverable to me (in a major city, FWIW).

So I went up to a couple salespeople and asked why. They couldn't say why, just that they did have it in stock at this warehouse and they themselves could ship it for X fee.

I don't really get it. They said maybe they were out of stock at the factory or online order fulfillment center. But the pages said they were available.

I'd imagine anyone concerned about efficiancy at Ikea would just order online.

The value proposition of the long Ikea walks is the set up rooms showing how the products look in use, allowing you to visualise before you buy. I take a friend and a coffee and take my time at Ikea.

> Grocery stores put common items like milk way in the back, in order to have you traverse the whole store and get exposed to as many meters of shelf as possible. They generally also like to change placement of product, to make sure you don't learn the layout and get too efficient at navigating.

Wonder if the covid-catalyzed uptake on grocery pickup options will change this in any way. I certainly discovered the joys of making grocery errands much more efficient, to the point where the decided majority of my store runs are pickup-only: I don't waste time navigating the store.

(And hey, the store can show me random products based off ML analysis of my purchase history, win-win.)

This just moves the tracking, stalking & dark patterns online.

Obviously they can't analyze your purchase history in-store, but I do like how sometimes grocery stores put commonly bought items together. For example the supermarket I often go to has marshmallows, Hershey's chocolate, and graham crackers together. They also keep cans of beans (and perhaps other stuff that goes in chili? I can't remember) by the ground beef.

> They generally also like to change placement of product, to make sure you don't learn the layout and get too efficient at navigating.

I dont know if this is really happening. Grocery stores where I go are fairly static. They change layout once in years and larger one have big signs that tell you where what roughly is.

Hotels don’t make money if you stay in the room, it actually costs them money in terms of electricity. But they do want you to spend money so they might indeed do things such as banning outside food deliveries or making the basin un-pluggable so that you can’t do your own laundry.

There was a case a few years ago where a Marriott got fined for blocking outside wifi.

https://www.theguardian.com/us-news/2014/oct/03/marriott-fin...

>Hotels don’t make money if you stay in the room, it actually costs them money in terms of electricity.

No, them renting the room to you - when you'll usually be out of it most of the day - is absolutely how they make money. Maybe in Las Vegas the room rate is a loss leader but your average business hotel is 100% making their profit on the room.

That's my kind of hotel, just a business that makes their money on exactly what the fuck it looks like they're making the money on.

I once stayed at a place that had super, super cheap refreshment rates on the mini-fridge. Cheaper if you got them at a store obviously, but not punishing like most other hotels, less than double the supermarket rate. Guess what? I made them tons of money, it was super convenient, I didn't go out of my way to avoid them.

What I mean is the hotel doesn’t make more money if you stay inside the room, which was GP’s premise that they would want to entice guests into spending more hours in the room. They would prefer that guests leave the room and go to the bar or restaurant for instance.

This is funny but on a side note, isn't this exactly what some casinos and some malls/stores do? I once spent one hour in a duty free mall because it was like a maze and I couldn't find the exit.

With reverse chronological timeline, it's harder to build a advertising profile.

While it's important to know what tweets a user is interacting with, the personalization option can surface more things as "You might like", "You follow x, and he liked", "Interesting tweet on the topics you follow/might like" and so on. Whether or not you choose to interact with anyone or more of those is a signal, and helps with advertising (if advertising is a question of "What tweet/content you are likely to interact with?"). Other than brands, most twitter influencers, or would-be influencers end up spending money on twitter ads to gain followers.

With a reverse chronological timeline, it's hard to surface those tweets from outside the network without a jarring experience. They also realize the power of defaults and hence they decided to take away the default option.

As a side-note: a good test of when social networks turn the corner and become evil is when they start conflating liking and sharing by implementing the "You follow x, and he liked" feature that you described. Since this clearly violates the user's intention (if you wanted to share something, you would have shared it), this serves as a subtle but distinct signal that manipulating users has become more important than listening to them.

> a good test of when social networks turn the corner and become evil is

... when they are created.

Social networks are inherently evil, that's their business model: manipulate users into addiction and serve ads.

How do non-evil social networks look like? They're forums, plain old forums.

But you don't become an ultra trillionaire with PhpBB.

When I joined FB there were no ads and a chronological feed and everything was awesome. Reddit was awesome.

Nowadays they are an evil manipulation mess, but I don't think it is an inherit property of social networks.

Phase 1: user acquisition.

Phase 2: profit (through using and abusing their users).

Both steps are written in stone in their business plan on day 1, you just don't know it.

As a user you were still in phase 1.

Similar story with image hosting websites. First they don't have ads and load fast and don't really have image restrictions. Then they start adding restrictions and ads.

Profit doesn't have to be equivalent to abuse. The idea that you have to abuse your users to make a profit is a relatively new one pioneered by Silicon Valley scum thanks to anti-competitive practices and lax/non-existent regulation.

Before this, we had centuries of businesses who were sustainable by just charging a fair premium on the goods they sold to their customers, and both sides were happy enough with the transaction to continue.

Well, considering the context of this thread, are there any social networks not working based on the Silicon Valley model? I don't know any.

The phone company? You pay them a fee for access and that's it - they have no incentive to prioritize certain callers over others for "engagement"'s sake.

In this context social networks = social media = Facebook, Instagram, TikTok and their ilk, we can't really include AT&T in this conversation :-)

Why not? My point is that a communications network doesn't have to be hostile. It is possible to run a communications platform and get people to pay for it.

Back in the day? Maybe. Today? Yeah, sure you could create something and try to charge people, but most people are now used to free + ads.

You'll never get enough traction to break through to the mainstream, unless something radical changes.

i'd argue social networks only become evil if the want to grow revenue and earnings and more importantly keep growing them at some fixed rate y/y for too long (otherwise they get recategorized as value instead of growth and all the juicy multiples disappear from their stock grants...)

Or federated like the Fediverse applications, i.e. Mastodon. It's much harder to squeeze users because they can leave you without loosing their social network.

> Social networks are inherently evil, that's their business model

What's evil about Mastodon, Matrix, Pleroma, diaspora, etc?

The fact that if they're good they're not adopted, I guess?

I don't know anyone using any of those networks.

Disagree here - you can insert things into a chronological timeline just fine - it's no more jarring than an algorithmic one.

Nothing wrong with showing me, and clearly denoting, a tweet that I might like in between other tweets I browse.

Don't understand this black and white thinking.

I agree that's a possible reason. But I don't think it's specifically necessary for Twitter, especially the sort of power user who wants a chronological timeline.

Whether or not they show topic data for a particular tweet, they generally have it. I only use the chronological timeline, and I cleared out the interests list a year or two back, but they have built up a profile of hundreds of interests for me.

I suspect forcing the ranked timeline on people is all about a small portion of Twitter juicing their engagement metrics, with no grand cross-team strategy in play.

It's also because there's power in being able to influence when people see something. It's harder to spot when that occurs when your timeline isn't chronological.

Which is what pissed me off the most about TikTok for the brief period I used it.

They would just show me random shit basically irregardless of date.. and that's not at all what I wanted. I found it to be an new source- of sorts. I'm not saying a good one, but at least it was somewhat interesting to may see peoples perspective on some topic. But what I found is it would just show me stuff that could be months old, days old etc instead of prioritizing like.. top day things.

I haven't used it during the russia/ukraine time.. but I can't imagine how wildly inaccurate of a picture you would get if you are viewing stuff completely out of date order. Yeah they put the date on the video.. but that doesn't really help, everything is still all over the place.

For all Reddit's faults- I can at least go there, take a look at some of the top headlines in some subreddits.. and know if I see something it's recent unless I am looking otherwise.

I would bet there’s a phenomenon at work here where you can addict almost everybody for a short time by doing X but long term a few become hardcore addicts and most become alienated.

It’s hard to tell that your actions have changed the kind of user who is most engaged until your platform is destroyed and consists of users who barely pay attention and addicts that feed off of each other and not much else (Facebook).

Metrics lie and optimizing for certain performance metrics ends up optimizing for exploiting human weaknesses which is actually somewhat uncommon in the population, which explains the process of optimizing your platform into irrelevance.

> They've been battling the widespread preference for chronological timeline

Or the simpler explanation is that the chronological timeline isn't actually the preferred setting, outside of a small but vocal minority.

If the chronological timeline is somehow hurting their profitability, why would they offer it at all? I think the more obvious explanation is that very few people were actually using it according to their analytics, so the switch no longer occupies prime UI/UX real estate. If it was some sort of conspiracy, they would have just killed the feature.

> As the common belief goes, the "algorithmic" one is more profitable. I buy this, intuitively, because it seems natural that it'd be harder to sneak in something jarring if you were viewing a more traditional timeline.

I don't follow this logic. There's nothing stopping Twitter from inserting the same ads at the same frequency just because the Tweets are ordered differently.

> If the chronological timeline is somehow hurting their profitability, why would they offer it at all?

If I recall correctly, this was the default. Then they introduced the semi curated algorithmic as a feature. Then replaced chronological and got a ton of backlash. They've been playing a game of capitulation and direction ever since.

> There's nothing stopping Twitter from inserting the same ads at the same frequency just because the Tweets are ordered differently.

The chronological timeline is your connected accounts + ads. The algorithmic timeline is the same but also some "suggestions." Ostensibly this makes it harder to immediately separate ad from, say, an account your friend follows.

You've mentioned it a couple times - their analytics may support that fewer users are using the chronological sort. I wonder if it's because they are frustrated and moving off to alternative viewing portals to get the feed experience they like.

> Or the simpler explanation is that the chronological timeline isn't actually the preferred setting, outside of a small but vocal minority.

Is this a simple explanation for why the feature automatically turns itself off after a week or two?

> just because the Tweets are ordered differently

This isn't what the feature does. More than half of the content in the algorithmic timeline was not posted or retweeted by someone I follow.

>Or the simpler explanation is that the chronological timeline isn't actually the preferred setting, outside of a small but vocal minority.

you can justify a lot of things with "the silent majority doesn't mind it". the silent majority is tech-illiterate, and there are a lot of things they are indifferent to. if it wasn't for the loud minority, we'd be advancing into the dystopian future at much faster pace.

>I think the more obvious explanation is that very few people were actually using it according to their analytics, so the switch no longer occupies prime UI/UX real estate.

the overwhelming majority will always use whichever option is the default

>If it was some sort of conspiracy, they would have just killed the feature.

that will eventually happen, and when it does, I'm sure I'll read about how that is a good thing

If the feed is strictly chronological, then the user has a better chance of knowing when they've seen everything, and won't come back to the content until more time has passed.

With the algorithmic feed, you could refresh the page at any time and get a different personalized set of posts that make it feel like your feed is more active than it actually is.

I agree, I think Twitter’s argument that the Home feed is actually more interesting for most people is accurate.

I myself tend to stick on Home. However, there are times when I want to be on Latest, particularly if there’s an event on or developments are happening at a rapid pace.

The old UI seemed ideal to me. It gave people the choice to choose what their default was. Now, they’ve taken that away… to what end? Force pushing people over to Home who expressly don’t want it probably isn’t going to work.

That’s a possible explanation.

However I’ve heard of very few people that prefer the algorithmic feed on Twitter, might be a bubble effect though.

The big part is time spent I think. I see the same thing with Instagram: I have about 120 accounts that I follow, mostly friends of mine.

I check the feed once a day and after 5 minutes I’m done.

But I also know quite a lot of people that spend hours on Instagrams explore page and with reels.

With your own carefully curated list there is a limited amount of content you can consume, until you inevitably leave the app.

But as you can see with TikTok or Instagram people spend hours on algorithmic feeds.

Companies need to service the vocal minority. This isn't a factory at Ford, they are more than capable and resourced enough to deliver colors other than black.

> But I can't fathom why seemingly no effort has gone into monetizing a chronological timeline. Why is it's foregone conclusion that it won't work with advertising the way the algorithmic was works?

Operant conditioning and intermittent reinforcement are why.

The chronological timeline is an algorithmic timeline. It uses an algorithm, sorting, to display the best content.

What they want to use instead is a different, inferior algorithm (for you, or you wouldn't hate it so much), but which is superior for them. That algorithm is not the only profitable algorithm, it is one of many, your preferred system is also algorithmic. This is just an algorithm that uses a black box nobody understands, or even tries to understand, or even tries to read (which they could, they have access to the weights of the neural net), and consequently, which they can claim to have no responsibility or even power over.

They're actually more heuristics, not algorithms. As such, they have all kinds of failures, many intentional. Algorithms (at least in my book) are generally are air-tight, and verifiable. Like you chronological feed, if two items were out of order, you could see it and it would show the algorithm to be wrong. It would have to lie about the date (which you do see sometimes) to fool you.

These heuristics are really just the man behind the curtain, imposing his will in an algorithmically-amplified way. It is driven by a dual goal of profit and political ambitions, with the latter being weighted more highly.

Why is it a foregone conclusion that it's about straightforward monetization, or even monetization at all?

What if it's actually a way to hide specific topics or sentiments from the timeline? If the algorithm of what to show first is a black box, scales can be tipped just like Google does with search results.

If it's a pure chronological timeline, you either have to be blatant in not displaying the tweets you want to suppress, or let people see them.

With chronological timelines you either "run out" of content (because you've already refreshed like 15 times this hour and read everything the people you follow have said), or you get buried in content you don't want (because you're following people for hairdressing tips and they're all tweeting 20-page threads about why Putin is actually in the right somehow). Eventually you burn out and churn out.

With algorithmic timelines the system is deliberately optimizing for "what keeps you on the site", even if it's not new or not something you follow. Hence, you stay on longer and view more ads. There's nothing about chrono that makes advertising less effective; algorithmic timelines just keep people on site and thus keep them as part of your ad inventory.

I think seeing everything from people you know and seeing things in chronological order are separate. Just because you select the latter doesn't mean you get the former.

Simple experiment:

Go to Facebook and use ?sk=h_chr scroll until you "run out" of material, and then check to see if you actually saw all of the recent stuff from a particular friend.

That is, not my experience with algorithmic at all... Algorithmic digs up days old posts I saw, commented and liked and slaps them at the front.. the new stuff, the stuff I'd want to see and interact with, just disappears.

But then maybe that's my fault for curating my timeline to show creators and not influencers/talking heads...

That's the point - algorithmic artificially lengthens the time you need to actually see new posts by making you scroll through lots of old bullshit - and you guessed it - ads in between and never giving you a clear indicator of when you've all caught up. By contrast, chronological will by default give you the latest stuff (which in case of news is what you want, and might even remove the need for you to see earlier stuff) and makes it super clear to tell when you've reached the end of the new stuff.

> Hence, you stay on longer and view more ads.

Jokes on them. I block every account that shows promoted tweets and it keeps my timeline ad-free.

I'm actually surprised they don't have promoted tweets override blocklists but hey, if it works.

Shhh, careful or they will catch on and my 10,385 blocks so far will be in vain.

A chronological timeline has benefits but will look very dull compared to a timeline where Twitter can put in front of you stuff that's related to recent news but that was posted a while ago. The key is to push topics in front of you, not the latest thing that was said.

The question is: who chooses the topic? In Google, the default results of the search engine can be circumvented by adding keywords to force the search algorithm to follow your thoughts. In Twitter, well... You hope Twitter won't be evil. [Guess what? It eventually will]

You're assuming that Twitter is a business that is run like a business and not something else.

None of those motivations seem to make much sense. I’d bet on simple negligence: only a tiny minority insists on the chronological timeline and it’s therefore acceptable to annoy them from time to time.

If I control your inputs, I control your outputs.

But why? I use it more now

Reply on news.ycombinator.com