A Database Of Shorts And Recommend Engine
LEO is undergoing a change that we have discussed for a long time.
Over the weekend, the testing of short videos on Leo was opened up. After more than a month of testing on the private network, it was pushed to the regular site. Here is where the community could test out the feature while providing bug feedback.
This was a major build. An entire back end was constructed to integrate with infrastructure that is not under the control of the Leo team. It is a tricky process meaning that diagnosis isn't always the easiest. That said, using the feedback tag to help point the developers in the right direction is netting some results.
There were over 20 fixes pushed yesterday alone according to what was stated on the AMA. This is something we are going to have to keep doing in an effort to get more stability.
In this article, we will discuss the power of this database along with the other features required for it to prospect.
Video Database ala TikTok
We all know how successful TikTok was. The idea of very short video content didn't occur to the major platforms until this application was on fire. It opened the door for a lot of copycats.
YouTube now offers them. Meta has reels. And Leo is adding it too.
Over the last year, we discussed the major problem with Leo, especially Threads, is the fact stuff gets buried rather quickly. There is little engagement with content since most live on the latest feeds. Anything over a day is basically in the abyss of the blockchain.
This is not the way to leverage the content of a database. Getting it filled is the first piece of the equation; being able to pull the information up is the other part.
It seems that we are lacking the second one within this ecosystem.
Perhaps shorts will be the first step in changing this.
For the moment, we are still operating under the old premise. shorts was added without the accompanying pieces. This was done for stability. The goal is to ensure this feature is working properly before adding.
Nevertheless, we can start the process of discussing what comes next.
Repurposing of Data
Look at this thread which has a short in it.
https://inleo.io/threads/view/ben.haase/re-cvvvfcpndm
This is something that was posted earlier today. Do you notice something about it?
We are dealing with a video that is timeless. In other words, people will equally enjoy it a year from NOW.
What this means is that anyone who comes across it, can hit play the same as today. There is no need to start afresh with a new video.
Consider what this is like one there are 10K videos of this nature within the database. That is what YouTube and others did so well. It is not only about content that was recently posted. They have years of uploads for people to view.
Which brings us to the additions.
For shorts to even be remotely successful more is required. We were told there will be a dedicated page for shorts, similar to what YouTube offers. This is helpful in that it will allow users to scroll down the page, viewing what interests them.
Another feature that is crucial is a recommend engine.
Online platforms excel because they have algorithms that curate content and place it in front of people. This means that people are given videos the application thinks will be of interest. We have seen this for years on YouTube, which might be one of the best at it.
It is through this that the database isn't wasted. Something that was posted 3 years ago could be of interest to someone, especially if it is timeless. This could be a song, motivational clip, or a speech. These types of things do not have an expiration date.
From what we understand, there will be a recommend engine also added to Leo. This will be for threads (which is what a short is). Using the actions of the account, it will do its best to match up content that it thinks might be preferred.
Of course, this will lead into another couple aspects of the overall Leo system. We will cover that in a future article.
Posted Using InLeo Alpha
I hope the algorithm that Leo will use on their shorts recommendation will be good. I think tiktok's algorithm was the one that really made it successful. It was showing videos that people really like and got even better over time. Another factor would be the type of content and users that Leo has.
A recommend engine is required. Without it, content gets lost forward.
I hope there will be a good recommendation system based on algorithm.
And for the rewards mechanism; old content can not be rewarded on Hive ecosystem. I wish there will be a Patreon style of reward system. So people don't have to upvote to give rewards, instead, support the creator with patron system with some benefits offer for the supporters.
There will be a subscription service added to Leo at some point.
We are dealing with a video that is timeless. In other words, people will equally enjoy it a year from NOW. - I like this idea , it's like a meal so good one had to save some for later.
That is where the videos from long ago on YouTube can keep getting views today.
Let's hope there will be a very good recommendation as it actually ought to be as that will help to improve a whole lot of things
Like any of them, the effectiveness of them is related to the amount of activity people do. Those who dont do much will find it lacking.