> The cost of hosting has remained steady over the past few years at $2.5million, despite traffic rising. The vast majority of the increase of the use of funds is for staff, most of whom do not contribute to the content or product in any meaningful way.
Do you understand how insanely difficult it is to operate a website at Wikipedia's scale, which has increased exponentially over time? The cost of hosting is negligible compared to the engineering effort needed to design & maintain systems that keep things functional. Here's a quick summary from their blog on the types of things they work on in a given month: https://blog.wikimedia.org/2014/10/18/engineering-report-aug... . If you think Wikipedia is just a simple deployment of MediaWiki sitting on a few DO instances, you are sorely mistaken.
Looking at the breakdown at 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/# 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?
> 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.
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?
Do you understand how insanely difficult it is to operate a website at Wikipedia's scale, which has increased exponentially over time? The cost of hosting is negligible compared to the engineering effort needed to design & maintain systems that keep things functional. Here's a quick summary from their blog on the types of things they work on in a given month: https://blog.wikimedia.org/2014/10/18/engineering-report-aug... . If you think Wikipedia is just a simple deployment of MediaWiki sitting on a few DO instances, you are sorely mistaken.