First, computer science relied heavily on models as well as algorithms before the advent of big data. People who wanted to get accurate conclusions needed to build models to describe the problem, and at the same time, they needed to rationalize the logic, understand the cause and effect, and design subtle algorithms to come up with conclusions that were close to reality. Therefore, whether a problem can be best solved depends on whether the modeling is reasonable or not, and the competition of various algorithms becomes the key to determine the success or failure. However, the advent of big data has revolutionized the reliance on modeling and algorithms. For example, suppose a problem is solved by Algorithm A and Algorithm B. When run on a small amount of data, Algorithm A produces significantly better results than Algorithm B. That is, Algorithm A produces better results in terms of the algorithms themselves; however, it has been found that when the amount of data is increasing, Algorithm B produces better results with a large amount of data than Algorithm A does with a small amount of data. This discovery led to a landmark revelation for both the discipline of computing and computer-derived disciplines: as data gets larger and larger, the data itself (rather than the algorithms and models used to study it) guarantees the validity of the results of data analysis. Even in the absence of precise algorithms, with enough data, one can get close to the truth. Data has thus been hailed as the new productivity.

Second, when there is enough data, it is possible to draw conclusions without understanding specific causal relationships.

For example, when Google helps users translate, it does not set up various grammar and translation rules. Instead, it utilizes the word usage habits of all users collected in Google's database to make comparative recommendations.Google examines the writing habits of all users and recommends the most commonly used and most frequently occurring translations to the user. In this process, the computer can not understand the logic of the problem, but when more and more data are recorded on user behavior, the computer can provide the most reliable results without understanding the logic of the problem. As you can see, massive amounts of data and the analytic tools to process them provide a whole new way of understanding the world.

Third, because of its ability to handle a wide range of data structures, Big Data is able to make the most of the data on human behavior recorded on the Internet. Before the emergence of big data, the data that computers can handle need to be structured in advance and recorded in the corresponding database. But big data technology for the structure of the data requirements are greatly reduced, people on the Internet left social information, geographic location information, behavioral habits information, preference information and other various dimensions of information can be processed in real time, three-dimensional and complete outline of each individual's various characteristics.

In the field of big data development earlier and do better is considered octopus collector.