In the Google search on "Alexa", Simplified Chinese the results of more than 70,000.    Now, "Alexa ranking" is the director of domestic websites are most commonly linked to mouth words.    Well, Alexa is doing this? What it came to domestic websites? Alexa What are little-known secret? This reporter through in-depth surveys and interviews, and work hard to restore a true readers Alexa, which is the first time a lot of information disclosed in the media. 

  Alexa betrayed 

  Wrote this subject, journalists too have defects, recalls carefully survey interviewed over the past few days to set up a Web site-impaired fans to the reporters to help friends, and journalists themselves, have invested a great deal of enthusiasm Alexa , in a sense, we are in the common "betraying Alexa." 

  In recent days, Beijing has just released a weekly magazine "China's commercial web site 100" list, of course, the list itself is not the "commercial" flavor, but to do so on some Web sites, the media has quoted another ranking data to illustrate the value of these Web sites, this was the "China Top 100 commercial websites," quoted the rankings list is the "world's leading third-party evaluation agencies - the United States Alexa" global website ranking.    Of course, the media quoted the Alexa rankings, simply to show that his "top 100 Chinese commercial web site," ranking authority.    And to another "authority" to establish their own "authority" is precisely shows that the media on the other "authoritative" - Alexa ranking recognition. 

  Alexa Who? 

  April 1996, Alexa founded in the United States, then it is only a small site, mainly Category navigation.    At that time, many sites are Yahoo Fans, Alexa is only one of numerous follow suit, the only one.    Alexa their own but there are also some technical means to visit a certain Web site traffic statistics and analysis.    Later, Alexa also to the development of the Internet users to provide their own search engine. 

  July 1997, Alexa is releasing a software, is now famous Alexa Toolbar (Alexa toolbar), which is embedded in a Microsoft IE browser tool, which each user access to the Web pages are Alexa back to a string of code, the relevant information, visit told Alexa. while in the toolbar users will see their Web sites in the global ranking of all website information, of course, this is only the website of Alexa rankings. 

  1999, the United States Alexa flagship e-commerce enterprises "Amazon" acquisition, a wholly owned subsidiary of the latter. 

  In the spring of 2002, Alexa abandoned its own search engine with Google instead.    The global reach of the Google Web crawlers constructed the pages of a huge database, which has greatly enriched the Alexa their own web site url database.    At the same time, Alexa also secretly released its own spiders, in the search for the unknown on the Internet web site.    After years of accumulation, Alexa URL of the Web site has been stored information at 40 billion, more than a search engine such as Google has become the Internet's most comprehensive Web site information database.    Alexa database daily average incremental up to 1 TB, and every two months will be able to carry out a comprehensive database update. 

  Now, Alexa toolbar in the world, "installed in the" tens of millions already, and Alexa main work is based on the worldwide desktop computer users the tools of the return of information on the Internet browsing habits global monitoring, and the development and sale of the kind of related products.    Of these, up to 100,000 Web sites worldwide web site list, for certain specific industries ranked site analysis, a customized web site for individual Web traffic monitoring reports.    Alexa also provided a lot of free Basic information, such as the "Global Site 500," and "100 Simplified Chinese website," Alexa Internet users can separate Show in a Web site's ranking on the history of changes in plans, and even at the same time the five largest Web site traffic and rank data intuitive horizontal contrast. 

  Although Alexa on its Web site to provide a lot of valuable information, but its real fame which is the four that were cited repeatedly by the media speculation has triggered a huge controversy and the global website ranking.    Alexa Site in accordance with the "official statement," Alexa open the former site of the world rankings by Alexa toolbar to collect all the information to return, after the summary rankings calculated.    From the "official statement", we find that the impact of Alexa rankings are two factors, first, Alexa collected information on these two is Alexa information processing method of calculation. 

  In the Alexa Web site, people can view a particular Web site to its own web site for the number of visits, Alexa provided a called Reach per million users (number of visits per million users, or Reach) data to show that the number of visitors, this data referring to the 1 million each Alexa toolbar users visit a Web site daily average number.    For example, on November 3, 2004, google.com value of the Reach 178,500, that is to say, on that day, every 1 million users in the Alexa toolbar nearly 18 million people have visited google.com. Only visit number, a Web site was not adequately reflect users in the situation, Alexa also provides data to reflect another of the Internet users visited the site-specific utilization, and this is PageViews per user (number of pages per user visit, or PV) .    PV value of a site all day-to-day visit to the site of the Alexa toolbar users to browse the website of the total number of pages on average, and the same people repeat the same page in the browser only counted as one day.    Also on November 3, 2004 visit google.com data as an example, this day, the PV google.com value is 4.0, that is to say, Internet users are visiting google.com average browsing the website page 4 . 

  With Reach and PV these two data, Alexa be able to flow in line with the global web site ranking, and its ranking of the site known as the Traffic Rank (traffic ranking, or Rank), according to Alexa on its Web site explained, This decision is Reach Rank PV and the geometric mean, that is the product of the square root of two.    Clearly, a Web site Reach and the higher value of PV, Rank also higher.    Still google.com example, the PV value of 4.0 in general, the web site is not high, but because the users of Google's search engine is more, far higher than the normal value of Reach website makes google.com Rank reach the 3, that is to say, google.com on the global ranking is third.    And the world's highest site is yahoo.com, Reach and PV values in the global web site are relatively high. 

  Use and suspicion 

  Every minute, every global installed Alexa tool of the computer terminals will report to Alexa at the end of the visit on the Internet.    According to this information, Alexa web site every day on the global rankings do recalculations, that is to say, Alexa ranking is updated daily.    Because almost all the global web sites in the Alexa within the scope of monitoring, and the daily updated list seems too intuitive, and the public is always available data, makes Alexa rankings on many occasions is very sensitive. 

  Since 2003, the domestic Internet industry has begun to heat up, they have been mercilessly get rid of capital Site CEO again regain its former sense, but they do venture capitalists than in the last.    COM bubble astute many, the website to attract investment, we must first dispel the worries of capital, even if the city is already on the website, shareholders also need to give some Clear data to show that the value of their own.    At such a time, Alexa rankings were introduced to the country, although the website had never aroused the attention of everyone, but it provided this list was quickly shown enormous commercial value. 

  Some websites to investors in the business plan submitted at the same time, intentionally or unintentionally, to mention start their own Alexa rankings, in accordance with the logic of some people default, the market in 300 sites in the world always Pai than 1,000 outside the site more valuable.    Finally, in the "circle" of word of mouth, Alexa rankings on the Internet in China has become famous, and this time, the other side of the ocean has its Alexa Fans who know nothing about China. 

  When Alexa ranking began to be generally recognized by the industry, has finally started to use Alexa ranking was to create "value".    The director of a large number of personal websites have started to study Alexa ranking rules, the Internet began to spread all kinds of cheating against Alexa tools, many beginning to appear on the Web site on Alexa cheating methods of articles in various forums on the Alexa rankings started in the discussion.    For a time, when it was China's Internet industry, "Alexa Cyclones." 

  At the end of 2003, in view of this Alexa heat, some industry also begun to question the credibility of Alexa rankings, and exposed to many websites cheating "insider."    In fact, a considerable number of Internet users have expressed Alexa ranking suspect in the 2003's because most of the time, Alexa ranking on the world's third - and fourth site is the two Korean site, this explanation is Alexa South Korean Internet users Alexa toolbar installed in a higher proportion, Alexa Internet users in South Korea to be relatively high sampling data, which resulted in South Korea's two portal into the global top five. 

  However, it seems that the interpretation of Alexa Huomaohuohei because Alexa toolbar always in English only, and neither the Chinese version, not Korean, and if it was high penetration of the tool, Europe and the United States should bear the brunt.    Some Internet users believe that accordingly, South Korea site is cheating on upgrading the website ranking, even more people think that since there is no Alexa Asian countries to launch localized versions of the tool, the Asian countries should firmly resist the Alexa rankings. 

  In any case, from the pursuit of the Bangsha, the concerns of the industry on Alexa not by a little bit.    Enter 2004, there are still many sites in the head racking their brains to want to have their own web site Alexa rankings "do upward.", Which had previously been exposed Alexa gunners have also carries a variety of purposes to the shelling Alexa . holiday after 11 of this year, domestic websites on the Alexa rankings suddenly experienced a large-scale collective landslides, various rumors spread in Internet users in the beginning, a more popular argument is that the algorithm Alexa finally began to adjust to counterattack more China's cheating site, Alexa people this is seen as "self-defence", after all, a list of the biggest selling points for the survival of the website is it the fairness of the rankings. 

  Onlookers were unable to solve all kinds of rumors have doubts about the credibility of the Alexa, as domestic mainstream media has never been carried out against Alexa technology in-depth analysis and reporting, and in the following introduction, the reader can see, perhaps Korean site rankings Another reason high, and so-called Alexa cheating is not just some media reports had repeatedly refresh so simple, and, Alexa also has its own very sophisticated anti-fraud measures.    However, as mentioned in this article and cheating means and the Internet can be seen everywhere, "Cheating Collection" was essentially different from Therefore, the extent to which Alexa could not resist such common cheating methods will be difficult to say. 

  Alexa tool of DNA 

  In order to completely clear whether Alexa ranking credible, must be technically on the Alexa Web traffic monitoring global comprehensive anatomy, of course, Alexa has never published the technical details of their own, the reporter decided "He's Road, Oxfam Bishen "Since Alexa claim that their data is the source of the tool, the reporter decided to begin from the start of the crackers. 

  Commissioned by the reporters, well-known circle of Web technology experts Kobayashi, spent nearly a night time, the Alexa toolbar and the return of Alexa data to a detailed analysis, a lot of valuable first-hand information.    Alexa many years ago to research done on the Kobayashi, the most recent this analysis reveals some hidden Alexa more technical details. 

  Kobayashi told reporters that now the latest version of Alexa tool of the operation mechanism and there is no significant change in the past, whenever the user to Alexa toolbar with the IE browser to open a new page, the Alexa a server (data.alexa . com) will receive encrypted data packets, this packet of information is the core of a dozen parameters, these parameters including the current web site, open the pages of time, the client Display resolution, the Alexa toolbar version, the user whether the "Amazon" users, an important parameters implied by Kobayashi analysis, that has been installed for each Alexa toolbar automatically generated id number, this number should be the only global.    Alexa can use the ID of each data packet sent feedback conducted only logo, which is PV solve problems and prevent duplication of the same users repeatedly refresh an important means of cheating. 

  Kobayashi told reporters that the results from the current study, a compilation of any expert can be very easy to master Alexa toolbar to return to data packets in secret, if this person is also a network programming expert, we should be for cheating on Alexa a relatively easier task.    Later, a reporter from the Shanghai Alexa cheating expert interviews, Kobayashi analysis is entirely correct.    The cheating, the Web is a senior development engineer, and the means of the analysis Kobayashi basic anastomosis - Alexa toolbar to prepare a return code generator, Alexa quantities to produce identification code string, and then use the multi-user virtual data.alexa.com the way back, and this can be deceiving Alexa servers, it is the misconception that these data to the different users (with the consent of the cheating, I agree with the topic in the final of the press and the public of the cheating Online Chat Record). 

  Kobayashi believes that this program to simulate multi-user access to the realization of cheating ways, the most important aspect is that the logo that Alexa only to the user ID, the identity of the Algorithm of crack, which required more than enough to Alexa Tools of a sniffer capture data packets quantitative analysis of the algorithm.    But Kobayashi also pointed out that these work in a programming expert, is not really what is the realization of the ultimate cheating still need a sufficiently fast simulation of generation ID procedures, which may be more difficult, but according to Kobayashi estimates, the domestic do these things minority people, the Internet is a master of this circle have only rarely do. 

  From the above technical analysis, Alexa day of the work done by the server is constantly receiving users worldwide send back data packets, which extracts more than 10 parameters, and then write specialized databases, and then in a specific time on the the same day to collect these data analysis and calculation, and a new calculation results to update on the website ranking.    The reporter observed that the results of the analysis of the database will be kept for at least three years, because the Alexa Web site for each site Rank rankings change trends up to three years to provide the data changes. 

  In the crack of the tools of DNA, the reporters also discussed with some of my friends in the Alexa toolbar global distribution.    Alexa according to the statement, it is the only tool of access to sources of information, then the tools of the global distribution of users will become another Alexa rankings can affect an important factor.    If the Alexa toolbar is really the only source of data, conceivably, when China's Internet users have not installed Alexa tool of the time, Sina portal Sohu such fear in the Alexa rankings are seen in shadow, but since the second half of this year, Sina and Sohu has followed Google ranked fourth in the world and fifth position, it seems that the Alexa toolbar has been in China, with a relatively high rate, otherwise, Sina, Sohu and followed the occupation of the Alexa Global 500, nearly 1 / 3 share of the China site on the suspects are cheating. 

  Fortunately, a friend of CAO Zheng-reporter is pleased to provide the data.    As a site traffic analysis experts, in www.tong123.com CAO Zheng-up to 2000 for various types of domestic websites to provide long-term traffic monitoring and statistical analysis services.    CAO Zheng-flow analysis and the different Alexa, Alexa was not monitoring the website do anything, and tong123.com pages on the web site the embedded form of third-party traffic statistics.    Commissioned by the press, in its interim CAO Zheng-sampling data analyzer, joined the Alexa toolbar on the monitor.    After a week of the statistical data, CAO Zheng-come to the conclusion that: Visit www.tong123.com systematic monitoring of the 2000 web site for all users, the Alexa toolbar in the installation rate of 1.5 percent. 

  Just released before CAO Zheng-back to the reporter on this to the results of the supplementary explanation, in his view, because the system uses tong123 cumulative average of the calculation method, Alexa toolbar proportion of the actual installation should be data than it is now higher because the monitoring of the newly added soon. 

  As tong123.com scope of the 2000 monitoring various websites, basically ruled out because of less subject to monitoring the user may choose tendencies and its credibility relatively high.    Even with the 1.5% rate of installation underestimated Alexa in the country to measure the impact of users has also been alarming, if the total number of domestic Internet users to 90 million, which will install the Alexa toolbar users may have more than 1.3 million .    According to Alexa claimed about 10 million downloads of the tools of the global total, the China Internet users Alexa enthusiasm seemed more prominent, it is perhaps also can be used as domestic websites this year in the list of overall top Alexa improve an explanation. 

  Since the tool of the distribution will be to a large extent from the impact of the monitoring results of Alexa, Alexa ranking may be a big geographical relevance, if this is true, Alexa authoritative really deserves suspicion.    Matter is not that simple, in the Alexa Web site, the reporter saw that the global web site before 100000 Price is the top 499 US dollars, it is clear that the other side of the ocean, and people still recognized Alexa ranking data, and even money to buy that Canada data . 

  Unwarranted privacy technology 

  Alexa in the various public information, or users of its discussion, it was not made clear Alexa toolbar in addition to any other outside access to the public network traffic monitoring tool, and a number of well-known websites for a long time to provide technical support Kobayashi, through the analysis of these web logs, there are many technologies that Alexa privacy.    Alexa could not be due to the official response, we can only say that these may be used technology is "groundless". 

  In order to further explore Alexa privacy, we have to first take a look at a Web browser process.    When the user opens the IE browser, in the address bar the importation of a string of web site and Enter, some http request contains data packets to be sent out, with other flows, these data packets to the user of the network gateway was routed to the public network, through a telecommunication room, and ultimately sent to a DNS provide analytic server, in the purpose of access to the site after the IP address can be put an ultimate aim of IP routing to the transmission path past.    Http request sent from the process, if Alexa all of the global DNS servers Sniffer analysis, it may be a fact that is very close to the global http request data, which can be very clear which sites were outlined what user attention. 

  However, to sniffer global http request, Alexa, almost impossible.    Withdrawal of 10,000 steps, even if Alexa can grab these data, it can not be the basis of these data to calculate capacity.    However, the Alexa rankings have shown some "jitter" signs seem to be telling people, in addition to the toolbar, Alexa is the use of technical means other suspects. 

  In the summer of 2004, Kobayashi found the website of Hong Kong, China ranked abnormal enhancement, such as "The Sun" this small site, even more than many mainland to the site in the rankings.    As telecommunications gateway and the system simplified and traditional reasons, the Mainland and Hong Kong Internet users are generally not "cross-regional exchanges," and the Hong Kong Internet users to install the Alexa toolbar will not absolute users than those in the Mainland.    The previous text of the analysis and calculation, the Alexa toolbar users in the installation of fear than the number of Internet users in Hong Kong are low, this is the case, the Hong Kong website Alexa ranking it is generally substantially elevated unjustifiable. 

  After two months of high-ranking period, the site of Hong Kong's ranking began to drop, but so far there are still some sites should be ranked higher than their actual place of the Department.    Kobayashi said that this phenomenon and Korea site in 2003 in a very similar situation, and South Korea's many websites are still high in the rankings.    If this phenomenon Alexa toolbar to explain the prevalence rate is unreasonable, but if the other by Alexa increase sampling machine to explain that, it is easy to say. 

  According to infer Kobayashi, Alexa may be set up in different parts of the world some sampling machine, with some telecom agencies or covert cooperation, or even at the approach of telecommunications gateway node network sniffer, get some "cover" sampling data.    The sampling data for the purpose of these is to take care of Alexa precisely that by the "tools of the installation of low-rate" in the sampling results.    Figuratively speaking, if Alexa Internet that South Korea has been very popular, and South Korea user has not installed Alexa toolbar habits, it may be to some place in South Korea Sampler, in South Korea to make up for sampling by the inadequacy of the tools.    The same situation may also lead Hong Kong has just set up sampling sites at the top of the surge.    After a period of observation, Alexa will be gradually adjusted in each region of the number of sampling, in order to get what it sees as the most reasonable outcome. 

  Kobayashi also of the view that even this year the website ranking upgrade of a large-scale sampling of a haunted, and the National Day of the top websites in the mainland after the general decline Perhaps it is because Alexa After a year of study, re-adjustment in the Chinese mainland proportion of the sampling results . 

  Of course, Alexa is to maintain its ranking authority and fairness, not only to improve the flow of data sampling process, but also to prevent cheating and make great efforts, those who tamper-resistant technology Alexa is not a secret propaganda.    Alexa then in the end is what methods to prevent cheating? In fact, the former reporter mentioned in the program to simulate Alexa toolbar to return data packet of cheating, Alexa almost no good way to guard against, or even detect the Cheating means it is more difficult. 

  In fact, the Internet is now common means of cheating, Alexa has already been sufficient immunity.    Generally speaking, now that the majority of search engines are not able to deal with JS script, focusing on search engines are often cheated by JS script prepared by the malicious code to achieve, and Alexa and Google adopted the spider is currently the only identifiable JS Script such procedures. 

  Kobayashi several large sites in the monitoring of the flow of discovery, Alexa opened a procedure called IA_Archiver robot, the robot procedures such as Google-like search engine spiders use of procedures, specialized crawling on the Internet, each Web page spying the flow of information.    Especially when the flow of more than a Web site Alexa set the threshold value, IA_Archiver will immediately climb up the site's server, the flow of this site and whether it is normal, no cheating.    According to Kobayashi IA_Archiver monitoring, the robot can be identified procedures have been most Web-based server-side flow cheating.    But now the industry on the general lack of awareness IA_Archiver reporter also in the Alexa Web site is a help page to see the robot IA_Archiver simple procedures, and internal procedures that the robot a few people, the lack of relevant technical studies. 

  Alexa trust? 

  Alexa rankings for the credibility of different people have different views on the Web in the Research and Application of the technology industry more inclined to such a statement: Alexa ranking is not necessarily absolutely accurate, but it is relatively credible.    Below reporters cite some examples from different aspects of the credibility of the Alexa rankings were compared. 

  Must be acknowledged that, despite a lot of technical means Alexa increased Web traffic monitoring data validity, but because of its inherent some technical flaws, it is inevitable that a small number of sites ranked resulted in a number of issues, although these problems to a large extent to blame on the website with the domain name of the irrational structure. 

  For example, Alexa ranking is the URL address, and do not consider the factors IP addresses, in large measure to reduce the Alexa ranking of the difficulty, but inevitably produce deviation.    Generally speaking, Alexa interested only two domain names, domain names for the three have been integrated into the flow of a domain name on the statistics, a relatively standardized major Web sites, it is completely correct strategy, but for a different to the user domain names circulated three sites, it should flow statistics will be all the users of the total flow? This obviously vary from case to case, a very clear example of such statistics is unreasonable cninfo.net I believe that the old domestic users still remember the domain name suffix, which is the provincial and municipal public CP two generic domain names, for example, and gd.cninfo.net sh.cninfo.net Shanghai and Guangdong are the two sites and Alexa reunification will be counted as www.cninfo.net of its subdomains.    Thus, www.cninfo.net became Simplified Chinese website ranked more than 40 sites, although it is not at all a matter of fact can visit the website. 

  Alexa rankings for another questioned the credibility of the different types of sites from the flow characteristics of the huge difference.    For example, portal and professional website or professional forum on the flow of almost not comparable because different types of users at different times on different occasions to visit the different types of Web sites, their browsing habits have great differences. 

  Even if the two sites according to Alexa algorithm that the comprehensive assessment of the same flow, it can not bring the influence of the two sites equated.    For example, a value of Reach 1000, PV value of a Web site with a value of 100 Reach, PV value for the 10 sites Rank should be the same, but the majority of cases, the content of the latter will be more valuable than the former .    Before a site visit because the users are only a page on Zoudiao, shows that the site may no attractiveness of a site, then the average user will see 10 pages left, the users are authorized The contents of the website.    Of course, the more extreme the situation will arise, for example, there is only one former pages, but the pages are very rich, while the latter are content to each page of little or user settings to a lot of unnecessary Jump, and even The latter might also be a series of novels station. 

  The existence of these complex people can hardly believe that Alexa Reach and can be used only PV hear all these values on the Internet site of the 369, etc..    However, if we look at Alexa from a different angle, we will find that its existence is justified, but also may indeed worthy of the trust of users. 

  If we have only the Alexa service as a traffic analysis tool, and its value has emerged.    Alexa Web site provides a very intuitive flow trend charts function, which in many other circumstances, we should pay to receive the service, and Alexa only as a free public service. 

  Alexa Figure 1 is provided by the Network of the World (ccw.com.cn) traffic rankings over the past year (Rank) trend map, from the map can clearly see that the site on November 3, 2003 to November 2004 on the 3rd ranking rose from about 7,500 to less than 2,000, and three of which were very sudden drop curve corresponding to the dates of the Spring Festival, May 1, 11 of these three holidays.    As we all know, the world of network users mainly concentrated in the IT industry, which led to the decline in three holiday curve very faithful record of the users in the time - because of a holiday, the majority of users to travel or rest at home, Internet users greatly reduced, reduce traffic, the site also will be ranked lower.    After the holidays, people return to the IT units, the first working day of the Federation of the Internet to look at the industry think the latest information.    So after the holiday, but the flow of ccw.com.cn holiday than before the upgrade to varying degrees, Rank reflected in the curve, that is the trough after three big curves with a high degree of restoration continues to rise. 

  In fact, if a detailed analysis of Web site traffic to changing trends, people can also distinguish between those who cheat site, Web traffic cheating because the changes are non-normal, and its website traffic and the normal curve certainly different.    Figures 2 and 3 were 265. Com dsdiy.com the past six months and the trend variation of the flow rankings, 265. Com is a Web site of China's famous navigation station, the last six months has been stable in the global rankings of 70 ~ 120 within.    Rank dsdiy.com the contrary, on the curve should be doubted, and the first week of September, the site's ranking on the global jumped from 100,000 to about 200, within one month later, the website has been in the rankings 100 ~ 400 hovering between, and the National Day holiday on the first day of the site's ranking has jumped to about 30, to the second week of October, the site's ranking quickly eased to 100,000 away in the Rank on the curve would never see, have similar rankings of the Web site and is also available sinapet.com haohz.com etc.. 

  Visit the Web as a monitoring tool, Alexa in the record to visit the web site of the true, but it also recorded the same site cheating those changes in the flow.    In this sense, Alexa is obviously a credible, and the key is that we must be able to choose the right perspective to play the role of Alexa. 

  No matter what kind for the purpose of selling Alexa always hope that people can be in the interests of the more in-depth study of Alexa, Alexa more profound understanding of the value, the more benefit from.