Baidu VS Google search ranking experience

– Google’s experience is better than Baidu, because the former rank is awesome

Where is the problem with this sentence.

First, the so-called “Google’s experience is better than Baidu” is a very subjective assertion and does not conform to the facts. If strictly speaking, it should be “for a specific group of people, for a specific field, Google’s experience is better than Baidu”.

In fact, as far as I can see, Google’s search quality is actually not as good as Baidu’s for many daily searches in China.

Second, in terms of the overall search experience, Baidu’s most criticized experience problem is the lax ad review and ads that confuse natural results. This problem is definitely not a problem with the rank algorithm. That is to say, the so-called “basically because the former rank is awesome” is also untenable.

Third, Baidu’s data search in the technical field, especially in some cutting-edge technology fields, is indeed a big gap with Google, but this is not mainly a problem of Rank strategy, but a problem of collection ability and collection direction. Google is a global index with richer information sources, while Baidu’s overseas index is lackluster, which causes two problems. One is that some high-quality technical resources are not included, which makes it impossible to give such search results; the other is that Although some high-quality technical resources were included by coincidence, various high-quality web resources that generated related citations were not included, resulting in a significant underestimation of their rank value, so high-quality resources could not get their due rankings.

Therefore, even in some areas where Baidu’s search quality is poor, the problem of insufficient indexing is actually a bigger component than the problem of rank strategy. The problem of collecting overseas high-quality resources is a dilemma that all domestic search engines have to face. It is really not a problem of money and resources, so I won’t discuss this.

Well, the above are actually some clarifications on search quality issues. If it is just these, this article can only be regarded as saliva, and the following points are valuable.

Compared with Google’s rank strategy, Baidu’s rank strategy is not as good as Google’s according to my observation, that is, Baidu is too indulgent to “click to raise rights”. Of course, this is my observation a few years ago, and it has been used in the past two years. Not much, so not sure if the status quo remains.

Well, my old article actually mentioned “click to raise rights” many times, but I didn’t explain it systematically. Why not? This thing is the job of many people, and even some people in my circle of friends. There is no reason to break people’s jobs, it is not suitable. Why do you dare to say it now?Because time has passed for a long time, the risk control strategy of many platforms has also gone up, so it is not hidden, and it is a case to enhance cognition for everyone.

A few years ago, some people criticized why Baidu searched for some common words, there would be very soft pornographic results, then some people would question Baidu’s integrity, in fact, they thought too much, this problem is caused by “click to raise rights” to escalate.

The “click to raise rights” is actually very simple. Under a certain keyword, the higher the user’s click-through rate, the higher the content weight will be considered. The starting point is to optimize the sorting of related results based on the user’s choice. A typical scenario of the machine learning of , if you only look at the logic of the original intention, it seems very reasonable, but soft porn content often has an advantage in the click rate under any word, so this situation occurs.

So the original intention of the algorithm itself is one thing, the complexity of reality is another.

Then the key point comes, if the search ranking can be changed based on the user’s choice, then what if the user’s choice is artificial?

As we said before, Google seems to have some control over click escalation, but what about outside of Google?

For example, the ASO industry is based on such a basic principle. Third-party ranking companies have at least several hundred million yuan in industry revenue each year. The logic is very simple. Use more different terminals to click on specific results under specific keywords to improve their rankings.

At first, it was a fleet operation, but it was easy to be recognized by the system risk control, but a new way of playing appeared, subcontracting, create a subcontracting task group, let everyone search for specific words, and click the specific results below to install. Earn commissions for assignments through video recording. This mode of risk control identification is very difficult, because they are all real users, and the time, location, and terminal type of the occurrence are also very scattered.

Is there anything else besides ASO? There must be, not to mention anything else, why many people on e-commerce platforms will brush orders, it is almost the same reason.

Then you say, why do you still need to use click to raise rights? In fact, the so-called recommendation algorithm isIt’s kind of a click right. It is based on the user’s choice, based on the user’s click, and based on the user’s download to change the weight of content distribution , so this gameplay not only affects search, classification lists, but even advertising.

Many advertising delivery systems make predictions on the quality of the advertising content, so as to give the delivery scale based on the prediction, especially for some advertising materials based on conversion pricing, the conversion statistics are delayed after delivery, and there is a fraudulent advertising system here time difference, then the question is, how to deceive it? It is to make the advertising system mistakenly think that this is an excellent material with a high conversion rate, so even a low unit price can give a high enough playback volume. How to make the system think that this is an excellent material, that is, when it is placed on a small scale, the conversion efficiency is artificially improved.

So this has become outrageous. In the past, we said that malicious clicks are clicks and conversions of competitors’ products, which cause budget losses to competitors. Now, in many scenarios, things are reversed, and they will consume their own budget first, and in the case of small scale to create their own high transformation illusion, and then increases the budget, deceives the delivery platform, and defrauds the high delivery scale with low unit price.

Deception algorithm is a big topic, and it is also a topic that risk control must pay great attention to the original intention of machine learning is to optimize the display weight of content based on human choice, but this logic will also be used deliberately. After all, ranking and push are behind the huge commercial value. And the related industry, to this day, is very amazing.

source：Google Chrome