Comment from the inside of Bing (Score 4, Interesting) 129
Bing came during the dark phase of MS- during the rule of SteveB. Steve put a culture in MS which was very toxic and obnoxious. Stack ranking had made people self centric, and they we not willing to help their peers, often obfuscated the code, and people went to the extreme of code stealing of others, and making it run under their names. Buttering the managers was common, and "I against you, us against them, and so on" was the mantra.
Bing had no answer to "why choose Bing over Google". During the all hands, this question was asked several times, but never answered satisfactorily by any one. DuckDuckGo figured out the answer- because they redefined user privacy. Bing had no such philosophy because they were blatantly copying Google.
In Bing, people from Yahoo India had come, with their stale ideas and the environment was completely political inside. Instead of writing code, and solving the issue of query share gain in a logical and original way, people generally were asking- "Look what does G show on SERP. Let's try and copy the result". As soon as a new uncharted territory (aka sub-vertical of queries) was found, the whole team would pounce upon it and try to hijack the project.
Even, some women (not all) were promoted because they had illicit affair with their program managers, and that was evident. Often, they would be called inside the chamber, and they would sit beneath the desk, and would be presenting the demo to him. These type of women were promoted on techniques which were shipped, and rolled back within 6 months because the issue ceased to exist, and some other team changed their web pages to more Bing-friendly way.
It was this era which saw exodus of brainy people, who did not want toxicity in their lives. In words of an ex- Principal engineer, "I was not able to make a dent in MS, but in the new company, whatever I do, I see a direct impact happening". The new company he joined was also in the league of giants.
The technology part was not matured inside Bing to be of any use then. The expressive power of language inside Bing was not sufficient enough to tell the ranking algorithm about the exceptions to be considered. As a result, Bing ranker would not work well on a generic playing field like internet.
Another issue that plagued the Bing team was query share distribution. we are talking which type of query has what percentage. Given that Bing used to get only 25% of the traffic volume of Google, the query share used to unrealistic, and most common question used to be- "as per our queries, what is the query share gain?" even though common sense told that queries of particular type should be prevalent.
Next issue was using unnecessary jargon to obfuscate the failures in training various rankers, or finding the query share. The jargon would usually make the talk convoluted, and cumbersome, and the main crux would be lost. Often, the discussions wold stretch hours, with almost no concrete plan to resolve the issue, but instead, tentative exploratory plans, which usually led nowhere.
managers had no clue about machine learning or artificial intelligence. they wanted a series of if-else statements, if that's what took to bring out results in a month's time. they had no time to train a model, test its scores on different metrics, and then ship. they just wanted quick outcomes, which seldom happens in the realm of AI / ML. There were no Ph.Ds initially in the engineering team, neither were any architects or Individual contributors giving design to teams to work upon. People were on their own.
Due to above scenario, usually people who got promoted were either proved loyalists, or the politicking ones. In any case, meritocracy was tossed out of window.
such a culture reflected the nature of top management, which was being copied by their juniors.
With Satya Nadella at helm, stack ranking was abolished for good, and the culture of trust was reinforced. People were pushed now to show results and were held accountable irrespective of the jargon they used, or the hype they created around themselves. However, Bing took too long to come in profit, and that too, too little.
Bing had no answer to "why choose Bing over Google". During the all hands, this question was asked several times, but never answered satisfactorily by any one. DuckDuckGo figured out the answer- because they redefined user privacy. Bing had no such philosophy because they were blatantly copying Google.
In Bing, people from Yahoo India had come, with their stale ideas and the environment was completely political inside. Instead of writing code, and solving the issue of query share gain in a logical and original way, people generally were asking- "Look what does G show on SERP. Let's try and copy the result". As soon as a new uncharted territory (aka sub-vertical of queries) was found, the whole team would pounce upon it and try to hijack the project.
Even, some women (not all) were promoted because they had illicit affair with their program managers, and that was evident. Often, they would be called inside the chamber, and they would sit beneath the desk, and would be presenting the demo to him. These type of women were promoted on techniques which were shipped, and rolled back within 6 months because the issue ceased to exist, and some other team changed their web pages to more Bing-friendly way.
It was this era which saw exodus of brainy people, who did not want toxicity in their lives. In words of an ex- Principal engineer, "I was not able to make a dent in MS, but in the new company, whatever I do, I see a direct impact happening". The new company he joined was also in the league of giants.
The technology part was not matured inside Bing to be of any use then. The expressive power of language inside Bing was not sufficient enough to tell the ranking algorithm about the exceptions to be considered. As a result, Bing ranker would not work well on a generic playing field like internet.
Another issue that plagued the Bing team was query share distribution. we are talking which type of query has what percentage. Given that Bing used to get only 25% of the traffic volume of Google, the query share used to unrealistic, and most common question used to be- "as per our queries, what is the query share gain?" even though common sense told that queries of particular type should be prevalent.
Next issue was using unnecessary jargon to obfuscate the failures in training various rankers, or finding the query share. The jargon would usually make the talk convoluted, and cumbersome, and the main crux would be lost. Often, the discussions wold stretch hours, with almost no concrete plan to resolve the issue, but instead, tentative exploratory plans, which usually led nowhere.
managers had no clue about machine learning or artificial intelligence. they wanted a series of if-else statements, if that's what took to bring out results in a month's time. they had no time to train a model, test its scores on different metrics, and then ship. they just wanted quick outcomes, which seldom happens in the realm of AI / ML. There were no Ph.Ds initially in the engineering team, neither were any architects or Individual contributors giving design to teams to work upon. People were on their own.
Due to above scenario, usually people who got promoted were either proved loyalists, or the politicking ones. In any case, meritocracy was tossed out of window.
such a culture reflected the nature of top management, which was being copied by their juniors.
With Satya Nadella at helm, stack ranking was abolished for good, and the culture of trust was reinforced. People were pushed now to show results and were held accountable irrespective of the jargon they used, or the hype they created around themselves. However, Bing took too long to come in profit, and that too, too little.
