Original link: http://chinese.catchen.me/2023/03/human-curation-vs-algorithmic-ranking.html
Why, as an internal communication tool, Slack works so well when the company is small, but becomes so ineffective as the company grows bigger? I have been thinking about this problem for a while, and I think it is a problem of data scale. Slack is a good solution when the scale is small, but once the scale reaches a certain level, Slack is no longer efficient.
If you’re comparing it to Slack, Facebook Workplace is very efficient internally at Facebook. Workplace has its own problems, such as causing employees to constantly swipe the newsfeed, because they are always afraid that they will miss some important information, so they will swipe every time they have time. Other than that, using Workplace within Facebook is a great experience, and important information always appears on the newsfeed. Because Facebook knows who you are, which colleagues you interact with more, and which content they interact with more, Workplace’s sorting algorithm can limit the content that is important to you.
Why is there such a difference between Slack and Facebook Workplace? The result of my thinking is that the difference between Slack and Workplace is essentially like the difference between Yahoo as a portal and Google. The former relies on human editing, while the latter uses algorithmic sorting. When the size of the data is limited, human editing can handle it and it can be very effective. After all, it is a person who reviews the content and decides what is important, what is not important, how to categorize it, etc. As the size of the data becomes larger and larger, human editors will eventually be unable to cope, which is why Yahoo will eventually lose out to Google as a portal site, and it is also why Slack is not easy to use.
Faced with a large amount of unorganized information, we have two very different options:
- Try to control the generation and flow of this information.
- Relinquish control over the generation and flow of information.
These two different directions lead to very different product settings.
The difference between portals and search engines can well explain how different the two directions are. Portals believe they can manually organize and index all the websites in the world, or at least the important ones. So if you’re looking for a certain website, you can either find it (indexed) or not (not indexed) on the portal. This is like a library. You don’t know if there is a book of a certain subject, but you can find the shelf where the book of this subject is located according to the book index. After browsing all the books on the shelf, you will know the book you want Does the book exist.
Search engines make a fundamentally different assumption. Following the example of the library, because the data scale is so large, even if you find the bookshelf where the book you want is based on the book index, you will find that there are so many bookshelves that fit this category that you can’t browse them at all. In a small library, computer science may be just a shelf. In the library of human knowledge, the shelf area of computer science may be several square kilometers in size. The premise of using a search engine is that you choose to voluntarily give up control of information. You ask the librarian to help you choose 10 books that are most relevant to the subject you want. Pick one that you think is most suitable and read it.
When using Google to search, although Google will tell you how many results there are in total and how many pages are displayed separately, most people don’t care at all, because the first few items have the information you want, and if there is no, then turn a few more times. page is also unlikely to have. This is what I define as relinquishing control: choosing to trust the algorithm to put the information that is important to you first, and forgo the information that comes later. No matter from which point you choose not to look at the information at the back, you can believe that you have browsed the highly relevant information, and the relevance of the information you do not read is definitely not as high as the information you have already seen.
Slack, email, and in-house human-edited portals are all part of the first option, which trusts people to control information. Slack can control the classification of information through channels, and can control the specific information you receive by joining and exiting channels. Regardless of whether you use Gmail or Outlook to receive emails, you can set complex rules to control emails, such as what emails are not read, and what emails are not read so frequently. The same is true for the company’s internal portal website. Editors often think about what information the company wants employees to obtain from the company’s standpoint, and then write articles and publish them to the corresponding channels.
Facebook Workplace falls under the second option. Using Workplace means that you give up manually (including by setting rules) to judge what information is important and what information is not. You choose to trust Facebook’s ranking algorithm to help you rank important information first, so no matter where you stop when you swipe newsfeed After that, you can believe that you have read the more important information, and the information that you don’t read at the back must not be as important as the previous ones. It is precisely because of this feature of Workplace that Workplace will be very useful in a company that is already very large like Facebook.
Today, no one uses portals anymore, everyone uses search engines, because the scale of information publicly available in the world is already very large. However, the situation faced by enterprises is different. Every enterprise starts from small and grows up. Slack will be very easy to use when they are still small. (I run a company with a few people myself, a Facebook Messenger group chat is enough, and I don’t even need to divide into multiple channels. Everyone knows all the discussions going on in the company, because there are not many discussions at all.) However, when the business Once you get big, it’s hard to switch to Workplace. Because of the essential difference in information architecture, it is impossible to import the information in Slack into Workplace, which makes Slack have great stickiness.
Workplace has some very successful multinational clients, such as Walmart and Starbucks. For a company of this size, using Workplace is a good fit. However, it is difficult for start-up companies to get off after they get on the boat of Slack. This has led to the fact that many start-up companies today need to find ways to deal with the various negative effects of Slack after they have reached a certain scale. I’d certainly like to be able to switch directly from Slack to Workplace, but really it’s a painful migration for any company of this size.
This article is transferred from: http://chinese.catchen.me/2023/03/human-curation-vs-algorithmic-ranking.html
This site is only for collection, and the copyright belongs to the original author.