How do they verify the person hitting "agree" is legally able to consent?
There's no way this holds up to the slightest challenge, but they probably won't have to face any challenge because who's going to do it? The math doesn't work for anyone and the legal system doesn't provide any allowance for "it's just harmful to society as a whole to allow this behavior".
Unfortunately for your argument, the brute fact is that courts often say otherwise if the company puts on sufficient evidence of how agreement is indicated and of its records indicating that the individual did in fact agree. See, e.g., Ruiz v. Moss Bros. Auto Group, Inc., 232 Cal. App. 4th 836, 844-45 (2014) (explaining how this works in practice) [0]; Aerotek, Inc. v. Boyd, 624 S.W.3d 199 (Tex. 2021) (same) [1].
Borrowing from what the late Charlie Munger famously said, you have to understand the incentives at work here:
• In the 1925 Federal Arbitration Act ("FAA"), Congress explicitly approved arbitration.
• Over time, judges have come to realize that arbitration agreements reduce their workloads: For a judge, it's much easier to review and (usually) "confirm," that is, enforce, an award by a neutral arbitrator, who conducted a fair hearing and allowed the parties a reasonable opportunity to put on evidence, than it is to conduct a full-blown court case, including overseeing the pre-trial proceedings and managing the complexities of a jury trial.
• Cutting the courts' workload even further: Under the FAA, a party that loses an arbitration has very limited grounds for appeal — the Supreme Court has said, in effect, you bargained for an arbitrator's decision in lieu of being able to go to court, so that's what you get. See Oxford Health Plans LLC v. Sutter, 569 U.S. 564, 133 S. Ct. 2064, 2068 (2013) [2]
(California and Texas allow expanded grounds of appeal if the parties so agree, but consumer arbitration provisions seldom include such agreements.)
• Another incentive: Because of the limited grounds of appeal of an arbitration award, a trial judge who merely confirms an arbitration award is less likely to be reversed by an appellate court than if the trial judge conducted a full-blown court case. That's significant because as a general proposition, trial judges are not fond of being reversed by appellate courts.
The above is adapted from the course materials (being revised) for the Contract Drafting course I teach for third-year law students. [3]
[0] https://scholar.google.com/scholar_case?case=751687725297847...
[1] https://scholar.google.com/scholar_case?case=617324454103183...
[2] https://scholar.google.com/scholar_case?case=129853902979909...
[3] https://toedtclassnotes.site44.com/Notes-on-Contract-Draftin...
In civil cases, on questions like this, courts go with the "more likely than not" standard of proof, as opposed to "beyond a reasonable doubt" as in criminal cases.
What courts generally don't do, once a party puts on sufficient evidence in support of its case, is speculate, without evidence, about alternatives that hypothetically could have occurred.
So: To start off, the company will try to put on evidence from which reasonable people could conclude that the consumer did in fact click on "I agree." (That's known as a prima facie case.)
But once that happens, as a practical matter it's on the consumer to refute the company's showing.
Sure, the consumer could claim that his dog or his kid clicked on "I agree," or that he spilled his drink on the keyboard. But the court will pay no attention to that claim unless the consumer can put on evidence to that effect, for example by getting on the witness stand and testifying under oath (and penalty of perjury) that he personally saw it happen. And then the consumer will of course be subject to cross-examination by the company's counsel.
I don't expect the law to at all times be perfectly consistent, but this seems to be a pretty blatant contradiction.
See my long response to @Brian_K_White just below (https://news.ycombinator.com/item?id=39650782).