Every time this issue comes up in an online forum, at least one commenter will pop up passionately defending this position. You can easily tell who it is in this submission (it's not you). The thing is, that's an economic perspective, but net neutrality is the name for a technological state of affairs. Focusing on the economic side of the issue subtly undermines the discussion by hiding the real problem before it can even be factored in.
Yet this is mainly a technology-oriented community (as far as I know), so let me try to convey my understanding of why internet pioneers, network engineers, startups and small businesses overwhelmingly support net neutrality (in addition to the afforementioned large online service providers such as Netflix or Youtube). Please bear with me if I write things you already know, and kindly correct any mistakes I make.
The Internet is a collection of thousands of smaller networks, many of which are international. It can only function as a whole because those networks peer with each other all over the globe and exchange information. This means John Doe, customer of ISP A in the US, can communicate with any other internet user - such as Hiro Nakamura, customer of ISP B in Japan, or Raj Singh, customer of ISP C in India. When a company or organization deploys their own network, they can join the internet by finding one or more peers who are already in it and entering a peering agreement with them. ISPs are classified into tiers based on the answer to the following two questions:
- Do I have to pay anyone for them to carry the traffic I generate?
- Does anyone pay me for me to carry the traffic they generate?
ISPs with a more valuable network - who own more actual long-distance capacity (bandwidth) in physical network infrastructure - are more likely to be paid and less likely to pay others, simply because they are providing infrastructure to others. But regardless of the economics of the peering agreement, usually the agreement entails carrying all traffic sent by the peer network to (presumably) recipients in your network, or your other peer networks. The traffic sent by the peer network doesn't necessarily originate in the peer network, but you treat it as if it did - this ability for traffic to traverse multiple networks that are not directly connected to each other is the only thing that ensures we have a single internet, rather than an endless amount of smaller international networks. This decades-old state of affairs, in which a network does not discriminate against traffic sent to it from another network by looking at where it came from, is called net neutrality.
When a company wants to be an ISP and sell internet access to the end user, they have to join the internet in the manner described above. After all, internet participation is always bidirectional - end users generate some traffic themselves. A new commercial ISP can join the internet simply by peering with a single existing network in a single location in such a way that the new ISP pays for all traffic they send over the network. In return, they can receive traffic from everyone else. They make a few access plans for the end consumer with as low an amount of upload bandwidth as they can get away with and price them in such a way that they'll be able to make a profit after paying their upstream provider and other operating costs. For redundancy and quality of service, they can peer with more networks at the same time and charge a little more.
But in the long run, as you can probably imagine, this is not cost effective. Assuming you are going to grow and have a very large participation in the internet (large geographic coverage, millions of users, lots of traffic), then you get a much better return by deploying your own network. By having a larger "network share" you are in a better position to negotiate peering agreements in your favor, not to mention all the traffic you carry within your own network doesn't have to be paid to a third party (or is subject to bandwidth limitations due to you only having purchased a limited amount of upstream bandwidth). Ultimately, an ISP like AT&T owns a backbone so large (currently 660000km of fiber according to their website), they're a substantial enough portion of the internet not to have to pay anyone for peering - it's mutually beneficial for them to peer for free with other large networks and make their money by charging smaller ones and having a bigger profit margin on their own customers, which makes them a so-called tier 1 network.
Enter american ISPs like Comcast. Comcast seems to have grown through a primarily end customer oriented strategy. They grew their last mile business very quickly and became a near monopoly for commercial internet access in many parts of the United States, serving millions of customers, all the while purchasing as little upstream bandwidth as they could get away with from larger networks with a strong presence in the United States, such as Level3 (now CenturyLink). This technologically short-sighted business model did not immediately fail because they "carefully oversold" their upstream capabilities (this is perfectly normal - you sell bandwidth based on your average network load expectations, not based on the assumption everyone will be using 100% of what you sell them at all times), engaged in p2p blocking and traffic shaping, and had a really good customer retention plan ("oh, you're not happy with your service? maybe you can try our competitors... wait, there are none? oops...")