Ideas like – from Mark Zuckerberg: ”Work fast and don’t be afraid to break things.” Overall, the idea to avoid working cautiously the entire year, delivering rock-solid code, but not much of it. A corollary: if you take the entire site down, it’s not the end of your career. – if we avoid semantics here these words are not encouraging us that performance analysis and testing are king roles there.

Again, I’m not interesting to bash anyone nor waste my time, but rather signal and be constructive and to do things properly means learning and looking outside the box we always like to live :)

The copy-paste idea here is that certain software pieces might or might not fit the business idea we want to build. But without testing and module validation, performance analysis that wont fly. These folks usually don’t have time to evaluate anything nor considering doing it. Because of certain other reasons: investors, time to deliver something, etc… They usually end-up “copy-pasting” things between each others (mileage might vary) but that’s the trend.

Example: this is how for instance MySQL got the attention vs. PostgreSQL, which has more mature and robust features than MySQL. Go figure.

As a positive note, I hope people will basically look and proper evaluate their needs before implementing something. Thats all what I’m trying to say. All the best

Some of the commonality in software stack across different companies takes place because everyone tries to leverage existing open source projects (faster/cheaper). In this case the system chosen may not be perfect/optimal – but it’s a lot better than starting from scratch.

Many of these choices are made early in a company’s life and are hard to change later on. A fast growing 1M user company will cut many corners and incur small inefficiencies to move fast. Those inefficiencies can become glaring and expensive once it reaches 400M users. But that doesn’t mean the original choices were wrong – simply that they need to be iterated upon.

One of the benefits of running a web site/service is that one can change the internals transparently. Being driven insane because of growth is a good problem to have (assuming that there’s an accompanying business model). One can always hire smart engineers and rejigger things (which is what Digg/FB etc. are doing).

All this said – it is pretty clear that Google has had an enormous impact on web architecture. Very few companies have the level of talent at the top that they do. As i was reading earlier today – good artists copy, great artists steal. You can call it cut’n'paste – but learning from the masters is not a bad thing at all.

Also confused about your copy-paste assertion. Each of those companies has developed — and released as open source — new components that don’t resemble anything released by the other two. In what way is that “simple copy-paste” with no regard to what’s suitable for their own needs? Would also love some examples here.

In short, I think you’re making a lot of accusations here. On what do you base them?

My point: “The Right Tool For The Job” is getting a thing
of the past for all these web wizards. Put your seat belts
on and wait 3-5 years more to see what these folks will create.

Common sense and things what we have nowadays somehow
manageable will be gone, extinct. Bloat dynamic sites with
data which nobody is interested in and massive deposit of
family pictures will dominate the future analysts press
conferences.

Sorry, I always look from a perspective of a sysadmin :(

Oracle shines in some areas where it has its problems somewhere else. What you refer as specialized hardware is something what you might want or not, again, depending of your business: regarding this majority of Oracle hdw is SPARC and x86 as any IBM or HP would offer as well. As well you mention closed software, Oracle has many sfw offerings including what it has received from SUN and majority of these
are not closed, instead are released as open-source projects: Java, OpenSolaris, GlassFish, etc. So again it depends…

I fully agree Nginx is a very capable server but for static content and it has its limitation handling dynamic content via FastCGI. Im using nginx and I really like it for its capabilities but millage might vary. I have evaluated a number of HTTP servers before selecting that – I did not copy paste what I read in Google ;) This is one of the majority mistakes many companies do.

To summarize all of these:

• There is a high percentage of copy-paste between Facebook, Twitter, Google and others. They dont even look whats good for them the simple copy-paste.

• The volume of data makes them drive insane. Can you delete your Facebook profile ? Do they recycle old data which nobody
  uses. Do they have any categorization what means old, obsolete data ?

• Poor testing: copy-paste adoption leads to poor testing
  and non-existent proof of concept phases where people suck
  things and digest hard. This makes them drive insane later.
  Proper planning and performance analysis is something which
  Facebook, Twitter and many other have no idea about.