But your general question is an open debate. If you run a hosting company do you have to host things you disagree with? If you run a transaction company doy you have to exchange with parties you disagree with? What if a certain type of transaction is too high risk, do they have to do it anyways?
Personally I'd like to see things lean more towards being agnostic to avoid amplification of the popular opinion but it isn't a cut and dry topic without corner cases or hard to define regulations.
So for instance this would be fine:
- A web hosting company refusing clients, because you can use another one
- Amazon refusing to sell the item themselves, because other companies can sell on the Amazon marketplace
- A bank refusing a client, since you can just use another bank
This would not be fine:
- Apple refusing to list an app on the app store, since no else can do so
- Amazon refusing to allow an item on its marketplace at all, since there is no other way to sell to people who buy on Amazon
- Google banning something from their search engine, since that's the only way to reach people who search with only Google
- PayPal refusing a client, since using PayPal is the only way to easily accept payment from PayPal account holders
The fringe cases I would worry about are for tiny niche markets where there is only one supplier, or regulatory monopolies (e.g. healthcare networks).
There are other app stores. Apple just doesn't support using them in their OS.
A much better example would be the ability to extend MobileSafari - Apple can and does do this (e.g. readability mode), but nobody else is allowed to (e.g. Instapaper). You can't choose a different default browser on iOS. There's a fine line between "not supported" and "anticompetitive".
Say you did get the regulation down perfectly though and it was dead simple to follow, why should a black rights activist have to do more work to find out which hosting company wouldn't discriminate while others can just click any and go? The popular debate extends to more than whether or not you think a certain position is enforceable as there is also disagreement about what reasonably solving the problem looks like. Some others would argue focusing on what the vast majority opinions are and ignoring the rest provides more value. Others argue it just provides the illusion that everything was solved instead of letting it be brought to mind.
(I should note that my politics is that of inherent distrust towards all large entities, whether they're governments or private entities. I would prefer the society and its economy to function on the principles that preclude either from existing at all. But regardless of that long-term goal, there's a question of what to do here and now, with the society and economy that we have. It's going to have monopolized markets - and proportionally sized governments are needed to be a check and balance on that.)
As to your second question - well, why shouldn't they? And how much is that overhead in practice, if we're talking about not picking one company out of many? The more socially unacceptable such discrimination is, the more attention to its violators - i.e. well-publicized blacklists, among other things. I seriously doubt that, in your particular example, a company that does that wouldn't be infamous for that exact reason.
And sure, there are people who would disagree with that. This is fundamentally an ethical question, and there's no single objectively correct answer to it; I merely gave mine. Some would say that anti-discrimination has to be enforced even in a truly competitive market, when there's no demonstrable economic harm, purely as a matter of principle. This kind of disagreement is meant to be resolved through a (hopefully) democratic political process.
But then, those laws literally got Borked:
Keep in mind that our modern anti-discrimination laws were passed at a time where that was demonstrably the case. And I'm not at all sure that it wouldn't also be the case if the anti-discrimination laws were repealed today. My point, rather, is that the purpose of those laws, and really any laws, should be prevention of harm, not a moral statement. If and when we get to the point where discrimination on some basis becomes so marginal that there's no measurable harm from it, it should no longer be illegal. But not until then.
Our present laws, for example, allow for organizations with private membership (e.g. churches, but also private clubs and similar) to discriminate more or less arbitrarily on their membership criteria, and then offer services exclusively to the members. It's only public services and offerings that are required to be non-discriminatory. Should we get rid of this arrangement?
Should this be expanded into the private sphere? Everyday activities include talking and otherwise interacting with private people, not just buying goods. If somebody decides to refuse to, say, greet people on the basis of their race or religion, should we consider such behavior inherently harmful, and prohibit it by law?
It's exceedingly difficult to function in 2019 on a strictly cash basis. Just like it's almost impossible to function in society without broadband. On the other hand, I can get by just fine if "hosting" doesn't agree with some questionable content I want to post online. It's more likely to get me a job than prevent me from getting one....
If so, you must accept, by law, all customers/participants/users without prejudice or discrimination, and you may only deny or limit service to same after a written order from a court of law instructing you to do so. Furthermore, this rule supersedes any Terms of Service that a company may wish to enforce.
That's not what we have in the US today—but we should. My money is on the EU crafting legislation to that effect first. I'd normally expect California to implement this kind of legislation first, but given the amount of money donated by the big tech companies, I doubt it'll happen.
 A "platform" is any business with a userbase that exceeds 1% of the population (for non-commercial customers, e.g. YouTube/Twitter/Facebook), or in the case of a business-oriented marketplace (such as Amazon's Seller Central), 10% of the specific market the business is targeting with their platform.
For example, if I host a very popular forum for discussing cars that somehow manages to exceed 1% of the population of the country it is offered in, I feel like I would be totally justified to remove posts that have nothing to do with cars and talk about completely off-topic stuff like videogames or politics.
That said, I see no issue with asking a privately run forum to keep it's membership under 3.4M if they don't want to be regulated like a platform.
It is in California (which, obviously, is located in the US).
Presumably landlords in DC can still kick people out for wearing an Obama / Trump shirt. (That’s supporting a candidate, not a party.)
I’d be curious to see court precedents on this.