> Open API's that reply in JSON
The serialization format doesn't matter. What matter is that we have way more APIs, all different, with inconsistent semantics and non-orthogonal feature sets than the non-web APIs of 15 years ago. On average the people designing these APIs are less competent than the people from 15 years ago creating more headaches for the people that are competent.
> Cloud VPS's at $0.02 per hour
Cloud virtualization is expensive, not cheap. It has its merits, the overhead for elasticity is much less than it used to be, but this matters only at scale. At small scale you get unpredictable performance and terrible I/O.
> 10 Gb ethernet
No such thing. Even on servers its rare. I had 1 GigE on my laptop 10 years ago. I can't get a better NIC on any laptop today. The world is actually worse than it used to be because back in the day I didn't need to worry about saturating Ethernet. Now I do.
> 54 Mb fiber in my house
I used to be able to buy a symmetric link with a fixed IP address and reverse DNS. Now fixed IP is a rarity, symmetric links are usually not available for non-business customers and when they are, they cost more than 15 years ago. Reverse DNS? Ha ha.
> multicore computers in everyones pocket
I don't need a multicore computer in my pocket. I need a phone with good signal and a battery that lasts. They don't make them anymore. Even if I need a computer, smartphones barely qualify. iOS is locked and Android requires me to do a type of programming I don't like. I'm used to computers that I can program the way I want, not being bound by some framework.
> GPS at everyones fingertips
I don't care. I never used a GPS, never needed one. What I've seen is that now people get lost when their GPS breaks. I view that as a failure of civilisation.
> web frameworks
I'd probably break some Hacker News scalability limit if I started writing about this one.
> caching solutions like Redis
Redis is required because the other pieces of the stack suck. It's a remedy, hardly a cure from an architectural point of view. The broad architectures around us are more unsuitable and more abused than they used to be.
It's actually worse than that. Unix and Plan 9 have thought us that's it is better to model behaviour through a single bounded interface rather than a growing set of specialised interfaces. This allows composition, protects against lock in, and allows synthetic components. Now there's a Redis API, there's a Cassandra API, there's a MongoDB API, there's a Zookeeper API, there's a Riak API, there's a RabbitMQ API. Everything has an API. A different API. Not only this destroys composability, it also hinders experimentation, increases the technical debt, makes the cost of transition higher, and bounds the writer into using a limited set of tools.
> data crunching pipelines like hadoop
Hadoop is a a player in an extremely niche field. I don't think it's relevant to talk about a thing as specific as hadoop in the context of something as general as the cultural and pragmatical shifts in the Internet. However, if you brought in the discussion, Hadoop is awful. Companies deploy it because it's trendy, not because they need it, introducing complexity, additional dependencies and a whole new set of problems to solve. Hadoop also dropped the bar on what is considered simple and sane deployment causing new software to be just as awful to deploy when they wouldn't really need to.
> payment processing like Dwolla
No idea what this is, but payments on the Internet are worse then they used to be. Sure, now you can buy anything, but it's harder to pay. Paypal periodically asks me for IDs and freezes my accounts just because I happen to move between two countries, there are many more types of cards, some work on the Internet better than others, some banks work on the Internet better than others. Merchants support only limited and disjoint set of payment options forcing me to have multiple types of credit cards and various types of accounts I don't want or care about. Back in the day, you had a credit card, it worked. Now I can buy groceries and shoes on the Internet. Back in the day I could not, but I didn't want to. I wanted to buy various types of equipement, and that I could.
> There will always be folks hankering for the glory days of alt.religion.kibology and compuserve. Ignore them.
The article was not about the olde glory days, it was about a fundamental shift in the way people and machines interact on the Internet. A transition from protocols to services. I think this is a worthy thing to discuss and your dismissive, condescending post is not warranted.