3 ms·

Could you quantize the exponential growth? The number of new pages is not growing exponentially (see http://www.wikistatistics.net/wiki/en/articles/full http:/

by dalke 12y ago

Could you quantize the exponential growth? The number of new pages is not growing exponentially (see http://www.wikistatistics.net/wiki/en/articles/full http://www.wikistatistics.net/wiki/en/articles/full and http://en.wikipedia.org/wiki/Wikipedia_article_traffic http://en.wikipedia.org/wiki/Wikipedia_article_traffic ). The number of active editors is decreasing (see http://www.technologyreview.com/featuredstory/520446/the-decline-of-wikipedia/ http://www.technologyreview.com/featuredstory/520446/the-dec... ). Looking at the breakdown at http://stats.wikimedia.org/EN/TablesWikipediaEN.htm http://stats.wikimedia.org/EN/TablesWikipediaEN.htm , and if I read it correctly, the doubling period for the number of articles is no faster than 6 years (2.4M in Jul 2008 and 4.7M now). http://reportcard.wmflabs.org/# http://reportcard.wmflabs.org/# suggests that the doubling period for the number of unique users is also no faster than 6 years, though that excludes mobile. By comparison, their IRS documents report $3.5 million in spending in 2007/2008 and $45 million for 2013/2014, which has a doubling time of about 1.5 years. I don't see how a doubling of services (or even quadrupling) across 6 years requires 10x more money. Could you explain why they don't have an economy of scale in their favor?

not_kurt_godel 12y ago
I was referring to traffic, not total number of articles or users (although both of those were clearly exponential for a number of years as well): https://en.wikipedia.org/wiki/Wikipedia:Awareness_statistics#mediaviewer/File:Alexa_traffic_ranking_for_wikipedia.png https://en.wikipedia.org/wiki/Wikipedia:Awareness_statistics... > I don't see how a doubling of services (or even quadrupling) across 6 years requires 10x more money Then you clearly have not done serious at-scale software development. > Could you explain why they don't have an economy of scale in their favor? Complexity increases exponentially (or worse) as traffic scales and the site matures, as you can clearly see in the blog post I linked. There is no 'economy of scale' when you simply need more engineers to manage that complexity.
- dalke 12y ago
  The graph you showed is from 2003 to 2006. At that time Wikipedia's expenses were something like $100K/year, and they had one employee. I don't see how it's relevant for this discussion. I searched for a similar graph for the last decade, but failed to find it. No, I haven't done at-scale software development. But every single report I've read about at-scale work says Google, Amazon, Wal-mart, etc. did not need 10x engineers in order to provide 2x content or 2x EC2 machines, at least not once they reached a certain threshold. Instead, the additional staff was used to provide more services. Given that it's true, I would like to know if it continues to be so. If increasing the Wikipedia traffic by another 2x causes the number of support staff to increase by another 10x then clearly it's unsustainable, and there is a serious problem in the immediate future. However, as the various reports on the topic have pointed out, at least part of the development cost have gone towards developing tools which had not had uptake by the Wikipedia developers. Hence why I would like to know what numbers you are using to draw your conclusion. What is the relationship between traffic and required engineering staff? Why doesn't it have the same cost savings that other organizations have reported?

by dalke 12y ago

not_kurt_godel 12y ago

I was referring to traffic, not total number of articles or users (although both of those were clearly exponential for a number of years as well): https://en.wikipedia.org/wiki/Wikipedia:Awareness_statistics#mediaviewer/File:Alexa_traffic_ranking_for_wikipedia.png https://en.wikipedia.org/wiki/Wikipedia:Awareness_statistics... > I don't see how a doubling of services (or even quadrupling) across 6 years requires 10x more money Then you clearly have not done serious at-scale software development. > Could you explain why they don't have an economy of scale in their favor? Complexity increases exponentially (or worse) as traffic scales and the site matures, as you can clearly see in the blog post I linked. There is no 'economy of scale' when you simply need more engineers to manage that complexity.

dalke 12y ago
The graph you showed is from 2003 to 2006. At that time Wikipedia's expenses were something like $100K/year, and they had one employee. I don't see how it's relevant for this discussion. I searched for a similar graph for the last decade, but failed to find it. No, I haven't done at-scale software development. But every single report I've read about at-scale work says Google, Amazon, Wal-mart, etc. did not need 10x engineers in order to provide 2x content or 2x EC2 machines, at least not once they reached a certain threshold. Instead, the additional staff was used to provide more services. Given that it's true, I would like to know if it continues to be so. If increasing the Wikipedia traffic by another 2x causes the number of support staff to increase by another 10x then clearly it's unsustainable, and there is a serious problem in the immediate future. However, as the various reports on the topic have pointed out, at least part of the development cost have gone towards developing tools which had not had uptake by the Wikipedia developers. Hence why I would like to know what numbers you are using to draw your conclusion. What is the relationship between traffic and required engineering staff? Why doesn't it have the same cost savings that other organizations have reported?