Huajiao live “Million Winner” added a “team to open black” function. Work in teams of four to challenge each other with the following rules:

  1. The user initiates a team invitation, friends can join by team number, and each person gets an extra resurrection card

  2. After forming a team successfully, users start to enter the program to answer questions

  3. If one member of the team passes the 12-question test, the team splits the prize money equally

Next, this article will analyze the realization mechanism and difficulty of team answer.

Now each question has 3 choices, suppose a person chooses the answer at random, under the million winner mechanism, he has 3 chances to be resurrected, the last question cannot be resurrected. His probability of completing the game is: 0.5 in 10,000.

Under the formation mechanism, each member of the small team has 4 chances to revive and answer the questions separately. Judging by the answer results, as long as someone answers correctly, the whole team will enter the next round. Team completion probability: 6 out of 10,000, 12 times higher. Simple math problem. If the number of people in the team increases to 6561, the probability of completing the game is 100 percent.

Technically speaking, you only need to add the following three points to the existing answering mechanism to achieve the team opening black:

  1. A team

  2. Synchronize the results in the team

  3. According to the team answer result, the team will enter the next round.

In essence, this is just a math problem that increases the probability, as opposed to traditional audio games where teams can hack and win as a team. From the operation point of view, is a new round of new activities. It’s not exactly a team hack.

What characteristics should team open black have?

  1. The team needs to be able to communicate. With only 10 seconds to answer the questions, it is impossible to communicate with words. At present the player’s own wisdom is, through the telephone connection open black.

  2. Have a mechanism for teams to submit answers. Because of the wisdom of the group, the group’s answer rate should be higher than the individual’s. The options are as follows:

    • After the voice communication, the captain submits his answer

    • The team members checked the answers separately and communicated with each other through voice communication. The team leader made the final decision according to the answers submitted by the team members

    • All team members can answer the question. The first answer is locked and can be modified, and the final answer can be submitted.

    • Group members communicate by voice, discuss answers and answer questions separately.

The difficulties and solutions of group voice hacking

1. Screen and answer questions should be synchronized between team members

If the pictures of the host are directly sent to the members of the team through CDN, the pictures of the members will be out of sync due to the uncontrollable CDN delay. Because the screen and the timing of the questions must be synchronized, this will cause the questions in the team to be out of sync.

Sonnet’s solution is slightly different. The pictures of the host are sent to sD-RTN, the real-time virtual communication network of sound network, through CDN, and then respectively to the members of the team. Due to the controllable delay of SD-RTN, and the delay is several hundred milliseconds, the pictures and answers of the members are synchronized, even if there is a slight difference between them, there is almost no perception.

2. The moderator’s speech serves as the VOICE communication between the BGM and the group and cannot affect each other.

Do not affect the BGM or allow the BGM to affect the call when the real-time voice is enabled or disabled or during a call. The App card should not drop or flash back, affecting the answer. And the volume should be smooth transition, can’t sound suddenly big suddenly small.

Soundnet provides a separate API to determine if someone is talking in the channel. When someone is talking, the volume of BGM can be automatically turned down; When you stop speaking, the BGM volume automatically increases. The whole process is smooth and natural.

3. Low delay for voice calls.

According to the current rule, the answer time is only 10 seconds. If the voice call is delayed, it will not only fail to communicate normally, but also affect the answer.

4. High concurrency

The high concurrency of live answering has its unique characteristics. When the time comes, nearly one million viewers will flood into a channel, which is a huge test for any service.

From the initial architecture design, acoustic network is oriented to mass and high concurrency. It uses global distribution servers and edge servers to access users nearby. When the backbone network is congested or even faulty, intelligent routing automatically avoids it, so as to ensure the high availability of services and the capacity of mass concurrency.

If you are interested in our solution, or you have any development problems, please visit our Agora q&A section to communicate with our engineers.