The fundamental problem of Western social networks is here: if you change it, you are seeking death; if you don’t change it, you are waiting for death (3)

Perhaps the real wealth is the friends we make along the way.

Shenyi Bureau is a compilation team under 36Kr. It focuses on technology, business, workplace, life and other fields, focusing on introducing foreign new technologies, new perspectives, and new trends.

Editor’s note: Why is there so much chaos in Western social media such as Facebook and Twitter today? What lessons should the next wave of social networking startups learn? Most of the best-known Western social apps choose to combine social graphs with news feeds. A product manager believes that using social graphs to generate interest graphs is the original sin of Western social media. The article is from the compilation. Due to length constraints, we have published it in three parts. This is the third part.

Image source: Pixabay

Highlights:

Instagram will start showing users content from accounts they don’t follow

Someone Some apps will use some form of theme or content selector

Many social networks do not have special design considerations when building graphs

The key is to think about what type of The social graph can provide the best user experience in the long run

In fact, the real wealth is the friends we make along the way

The fundamental problem with Western social networks is here: change If you are seeking death, you will be waiting for death if you don’t change.

The fundamental problem with Western social networks is here: If you are seeking death, if you are not changing, you will be waiting for death.

Recently, Instagram announced that it will start showing users messages from people they do not know. The content of the account you are following. In many ways, what we see from Instagram can be seen as

Compromising as much as possible to the superiority of TikTok’s pure entertainment architecture

Some apps will adopt Some form of theme or content selector. Tell us what kind of music or movies you like. What news topics interest you. They then try

to use machine learning and signals from the entire user base to provide relevant news feeds

The effectiveness of this varies widely. Why is it that Spotify can generate such a good playlist from just one song, but its podcast recommendations feel average? Why, after spending years and millions of dollars on research, do Netflix recommendations still feel so mediocre? And why do you say these are actually not important? Why are Amazon’s book recommendations reliable, but article recommendations on news sites feel random? An in-depth understanding of why certain content recommendations are so much better than others would require its own article because the subject is so complex.

But the focus of this article is on graph design, and the key point we emphasize is that things like content selectors should be explicitly kept away from the social graph. Twitter allows users to follow both accounts and topics, which can be seen as a half-step towards a pure interest graph.

This does not mean that apps cannot become more interesting when they are socialized, nor does it mean that people will not share things that everyone is interested in with people they know. We all care about our own interests and those of those around us.

When the interests of both parties coincide, the effect will be better

. It’s just that after more than a decade of using current social apps, there are now enough case studies to illustrate the point that the assumption that they are completely related is flawed.

The second thing to consider is what type of interaction the app will develop in the long run.

Is it a one-on-one interaction or is it broadcast to a large audience? How many of your users do you want to be both consuming and creating? What kind of map is best for your app? A map of people who know each other in real life? Or through a graph that connects strangers with similar interests? Or a mix of the two? Is your app intended for people from the same company or organization? Will interactions cross cultural and national boundaries, or is it best divided into separate maps based on different geographical areas?

Next-generation social product teams can and should be a little more proactive in thinking about what type of social graph will provide the best user experience in the long run.

I'm not sure about the answer, but based on the history I've heard, I think

Many social networks do not have special design considerations when building graphs as an exercise.

. This makes map design an exercise in which you face more open questions than answers. To the extent that Facebook was originally developed for Harvard students, this may happen to impose some useful graph design constraints.

Unlike some types of design, diagram design itself is not suitable for prototyping. Social networks are, at least in part, complex adaptive systems, so it is difficult to prototype what types of interactions will occur when the graph reaches a certain size.

But while traditional complex adaptive systems are so complex that all predictions are futile, social networks differ in two ways.

First, human nature is consistent. Second, we have many large-scale social networks to learn from.

These are plenty of real-world test cases that illustrate what happens when you make certain choices in your graph design.

They also exist in multiple markets around the world. This makes it possible to study different path dependencies, especially when comparing unique cultural and market conditions like China to the United States. Despite all the differences in context, problems like trolling appear to be widespread, suggesting that a certain Some effective underlying mechanisms are at work.

Once you get the hang of graphic design, there are a lot of rabbit holes you can dig down. How do you build enough trust when the people you're connecting to are complete strangers? If the backbone of the app is dynamic news, do these messages have to be specifically extracted from stories posted by accounts that users follow? Do we have to select candidates from these accounts? Is News Feed even the right architecture for healthy interactions between users?

Who should consider map design? When should you consider it? For example, the map design should inform the growth team strategy. Growth teams should not be viewed as rogue teams whose only job is to expand the graph in every possible direction. They need to know what good graph growth versus harmful graph growth looks like so they can develop strategies that are more aligned with their long-term vision.

Recently, TikTok has begun to push me to connect more with people I know in the real world. I got prompts from them asking me to follow people I might know, and now when I share a video with someone, I often get a notification telling me that they've watched the video I shared. These notifications are often the only way I know they have a TikTok account and what their username is.

So far, I’m enjoying TikTok and being able to see interesting content without having to follow anyone I know in real life. Perhaps TikTok is trying to make its video sharing endogenous to the app itself. But obviously, by now, I think any changes to the graph of any social product should be treated with more caution. Most people I know don’t make any TikToks themselves, so following these people won’t have much of an impact on my recommendation page. For young people, a higher proportion of this group of users make their own TikToks, so it may be more meaningful to follow each other.

On the other hand, as long as any app has a default public *** graph structure, it will exert the innate impulse of human judgment. Wait, what accounts on TikTok does this person I know follow? ! Tsk tsk.

Should TikTok urge users to copy real-world social graphs? The answer is not dead. I ask this question just to illustrate that graphic design is a subject that requires deeper thinking. As the name suggests, it can come in a certain design.

“Focus is a good word. Who we pay attention to can become a self-fulfilling prophecy. First you build your map, and then your map in turn shapes you. Numerous studies show that humans The people you spend the most time with tend to have the same rhythm. Silicon Valley legend Naval Ravikant has a very popular five chimpanzees theory, which is that in the field of zoology, you can tell where a chimpanzee follows. Five chimpanzees playing together can accurately predict the mood and behavior of that chimpanzee

The social media version of this theory is that we can tell who the user follows, who follows the user, and what they are forced to follow. The "space" where people interact is used to predict any user's behavior on the app, whether it is Facebook news, Twitter timeline or other structures. We all know that some people are their worst selves on social media. The fundamental attribution fallacy predicts that we will assume that they are inherently bad, without considering that they may simply be reacting to circumstances and stimuli.

Humans are not chimpanzees. We can often act as members of dozens of different social groups at the same time. Reed's law predicts that the utility of the network will grow exponentially, because not only everyone in the network can be connected to all other nodes, but the number of possible subgroups of the network can also reach 2^N-N-1, where N is this Number of people on the network.

But whether social applications allow such subgroups to easily form is a design problem. Integrated News Feeds tend to force people into larger subgroups, but that size is no longer conducive to healthy interaction. Although each user will see a different Twitter timeline or a different Facebook news feed, it will still give the illusion that this is a large public place. Because anyone can see what you post, you should do the same as everyone else.

In contrast, chat applications often allow users to form subgroups that are most relevant to them. The structure of FacebookGroups is more flexible than dynamic messages. People are divided into groups, and social applications should flexibly handle their various communication privacy needs.

It’s not surprising that soon after many technology companies installed Slack, they suddenly found themselves dealing with various employee riots. Once you rewire the communication topology of any group, you are changing the dynamics between members. Slack's public chat channel serves as the company's internal public forum, allowing more employees to understand each other's ideas. This can lead to employees discovering that others share opinions, such as reservations about a particular company policy, that they thought were the only ones they thought were shared by others. In the past, the reason why we have seen many companies appear calm is actually largely due to the inherent privacy of email as a communication technology. This privacy has concealed the surging undercurrent. .

In many ways, graphic design is more important to many current Western social media than to early social media, which is inevitable. In the early days of the Internet, public social graphs were sparse or non-existent. In most cases, our graphs are limited to the email addresses we know, and the username of someone on our favorite newsgroups. In the early days of the Internet, discovering a new online connection was like discovering an incredibly exciting secret, a feeling that is hard to explain to a generation that has grown up with the Internet.

At the time, it was very difficult to track someone online if you just knew their name.

Nowadays, we have enough ways to connect with almost anyone in the world. Adding someone to my address book is almost unnecessary when I have a dozen ways to reach anyone using my smartphone and the Internet.

In a world where finding people online has become a commodity, the better technique is to connect with the right people in the right environment. I have more than a dozen chat apps installed on my phone, and they all look roughly the same. While I discuss map design here primarily out of defensiveness—that is, to think about how to avoid map design mistakes—the positive perspective is to proactively use map design. How can a unique map be constructed so that its structure can absorb valuable and, more importantly, unique intelligence?

LinkedIn may be everyone’s favorite social app to complain about in Silicon Valley, and while many of the complaints are valid, the company’s huge market capitalization proves the value of its graph. It turns out that if you map a career, not just today, but across time and organizational dimensions, recruiters will spend a lot of money turning it upside down.

As for all the debate about whether our current social networks are beneficial to society, I prefer to focus on the potential we have yet to realize. Yes, we already have the wonders of Wikipedia, yes, but aren't there more types of large-scale collaboration that are possible?

Every week or so, I'm recommended to someone amazing, or an account I've never heard of before that blows my mind. The fact that social networks themselves don’t facilitate these recommendations doesn’t make me discouraged, but less sad. Ten years from now, today’s social graphs will look like blunt instruments, their configurations too primitive.

When we look back on those ten years, we will also see how many amazing people we met at the right time and in the right context, and realize that the real wealth is actually the friends we made along the way. friend.

Translator: boxi.