Is it because they are so good, there is no one better than them and therefore people keep voting for them for decades? Or is it because their respective party deliberately does not put forward anyone who has chances to win in the primaries against them? Or voters just vote for the party, that is for whoever the party puts forward as the candidate?
How does this work in the US?
This is almost entirely due to the party system we use in the US. Parties create the following effects:
People like McConnell, Feinstein, Lindsey Graham, and even good-ol' Joe Biden might have had a difficult time their first election cycle, but after that they were effectively shoo-ins: their own party wasn't likely to offer (risky) alternatives, and the opposing party wasn't likely to commit resources to (implausible) challengers. Such people effectively run unopposed on all subsequent ballots.
I can't say I approve of this type of political realpolitik, but it is what it is…
Generally speaking there's a large incumbency advantage. There's been a lot written about that. See e.g. this paper.
I'm quoting this bit because I find it particularly funny/insightful (even though it's not technically about Senators in the first part, although the paper makes that jump):
Even though Congress has low approval ratings overall, according to Gallup polls, almost half of Americans say they approve of the job their representative is doing. This implies that the public tends to blame other members of Congress for decisions they dislike, and thus opens the door for senator reelection. Drain the swamp, sure. But don’t throw out our senator.
But as that paper details, the same principles apply: name recognition, war chests, piggybacking on successes that they might not have contributed to much, etc.
More closely related to your Q though:
candidates with stronger chances of success – usually ones who have been in an elected office before – tend to run for open seats rather than challenging incumbents.
Alas that paper doesn't discuss that data in detail. There are some such papers though, e.g.
Studying U.S. state legislatures, we find strong evidence of strategic behavior by experienced challengers (consistent with previous studies). However, we also find that such behavior does not appear to significantly bias the estimated effect of challenger experience or the estimated incumbency advantage. More tentatively, using our estimates, we find that 30-40% of the incumbency advantage in state legislative races is the result of “scaring off” experienced challengers.
Alas that's not about the US Senate, so who knows how big the "scaring off" factor is in that case. OTOH what they did study was:
state senate elections, and measure challenger quality in terms of previous experience as a state representative.
(Using [11] state senates gives them a larger data set, so they're more likely to get statically significant, thus academically publishable results.)
Also, there's a paper that calculated that the incumbency advantage increases linearly with time in the US Senate. Their data set spanned 1952 to 2008.
And something in there about what I said about statistics and publication-worthiness (but also a list of other factors):
The most striking shortcoming of the incumbency literature is the lack of interest in senatorial elections. [...] The most common justifications used concern the higher quality of senatorial challengers (Adams and Squire 1997; Hinckley 1980; Lublin 1994; Stewart 1989; Squire 1992; Squire and Smith 1996), the difficulty in establishing a personal relationship with constituents in densely populated states (Binder, Maltzman, and Sigelman 1998; Hibbing and Alford 1990; Hibbing and Brandes 1983; Nice 1985; Oppenheimer 1996), and the higher level of media coverage and exposure that senatorial campaigns generally enjoy (Druckman 2005; Hess 1986; Sinclair 1990). In brief, virtually all these explanations can be traced to the fact that senatorial campaigns are held statewide. The unspoken truth, however, is that the Senate represents a less attractive statistical laboratory than the House: its members are fewer, they are elected less often, and they win fewer electoral contests.
Some incumbency effect appears in their 2nd term already, and only grows after that; some potential reasons are laid out:
The average share of the vote obtained by successful non-incumbent candidates is 55.4%. This number rises to 58.1% after six years or one term in power, 60.2% after two terms, 60.4% after three terms or 18 years, 62.4% after 4 terms, and 66.5% after five terms or 30 years.
[...] It might, thus, not only be true that sitting senators receive more votes at their second bid compared to their first bid: their vote share might also increase during their third, fourth, and fifth reelection. There are mainly two reasons why this rationale might be true. First, the longer a senator is in office, the more he or she can create bonds with his or her constituents. Second, the senators’ political clout increases with seniority. For example, seniority boosts a senator’s standing within his or her respective party and renders it more likely that the person will gain chairmanship of an important committee. Having spent many years in Washington, Senators are also more likely to gain other informal or formal leadership positions within the political establishment in the capital; factors that allow him or her to serve his or her constituents better by, for example, bringing home more “pork.”
Unfortunately, this paper doesn't empirically test the relevance of those putative factors. It does test a few others, but they're kinda "meh" and not in the [causal] direction of interest to this Q, e.g. it concludes that an incumbent senator whose election is contested by a counter-candidate with some elected office experience is more likely to lose their seat than otherwise [ceteris paribus], but this is a rather obvious fact... OTOH, we shouldn't totally discount the relevance of that empirical datum because it lends more credence to the idea that the "scaring off" effect of such counter-candidates [identified in other research] factually matters, i.e. such counter-candidates would indeed be more likely to unseat a Senator.
Electability of incumbents being higher goes a long way to explaining why people stay too long. Other answers seem to have covered that. There are other factors, though.
Which way a Congressperson votes and their party affiliation are among the most important things to politics, if not the most important, and so long as members vote the right way and have the right party there isn't much incentive to replace members. The competency required to just vote along with the party is frankly low, and as long as they do that, the incentives for the party to get rid of them are low.
There are things that make long-time members of Congress more effective. Mitch McConnell is undoubtedly an extremely effective senator, having been the majority leader most recently in 2021, so it's been surprising to have seen his recent lapses, but I haven't seen evidence that he's not still an effective senator, both for advancing his own party as well as getting legislation done with the Senate as a whole.
When you've been in the Senate as long as McConnell has, both Democrats and Republicans come to him to get deals done, because of his influence. This is both because he's been the majority and minority leader for the Republicans since 2007, and because of relationships he's established. More seniority means generally better odds of being in leadership, and once you're in leadership you tend to stay in, as long as you're performing the job.
There is an underdiscussed reason for why incumbents tend to stay in, though, and that's committee assignments. It tends to be that the most important committee assignments get first offered to the most senior members, and committees are powerful for a variety of reasons. In the Senate, the most senior senators of each party get offered the Appropriations Committee. If your senator has that much influence over the budget, why would you want to get rid of that? Here's an interesting article on when Senator Patrick Leahy was offered the head of Appropriations due to seniority, but declined in order to stay on Judiciary.
Simply because those with the most experience getting elected tend to be those who get elected the most. It's a purely Darwinian process. And it is not anything specific to the United States, nor to this particular field of endeavor, but is a universal and cuts across other fields and endeavors.
The "getting acquainted with the process" learning curve is, by itself, a significant barrier to entry. The learning curve of finding the best strategies, the best way to address people, fund-raising and so on, is something that favors the most experienced; not to mention name-recognition. The expression "whoever has will be given more" applies here.
The issue has been indirectly addressed, with a more global perspective, in another forum:
Who is the oldest parliamentarian (legislator) on record?
Who is the oldest parliamentarian (legislator) on record?
There is also a getting-acquainted-with curve for the electors. There is a strong tendency to go with the flow, regardless of the kind of political system, and that means: keep electing the same people over and over. That's related to the phenomenon of
Social Inertia
https://en.wikipedia.org/wiki/Social_inertia
and applies across the board - not just with humans, but with other species. So, if you study the hierarchical or leadership structures in other species, you may find the same thing happening: those who are in charge tend to stay in charge.
There was one reply that said "it is almost entirely due to the party system in the US" - which is nonsense. Because if that were the case, (particularly with the "almost entirely" attribute) it would only be true in the US and in other places with a "similar" party system, however you want to define that. That's how you test that proposition. But, I think it's pretty clear that this will not be the case, the test will be found wanting, and that the assessment was just a one taken from a rather naive, parochial, or even provincial, point of view, unaware of what's going on in the wider world.
This is also why you tend to see guilds and dynasties form in such fields as the film and arts industry. Wherever you have any kind of learning curve barrier to entry, then there will be built-in advantages for those households where the experience is deeply ingrained in the family as a form of collective knowledge and experience, passed down as part of the parenting, and where the off-spring see the process in motion from day to day.
(Other people have already brought up why these candidates are likely to keep winning primaries --incumbency advantages, support from party elites, etc.-- so I'll focus on why these candidates win general elections.)
Fundamentally, the reason Mitch McConnell is still a Senator is because he got more votes than his opponents.
"OK, but why?" Well, because most people in Kentucky liked Mitch McConnell better than the Democratic candidate.
"OK, but why?" Because Kentucky is a conservative state. Kentucks (can I call them that?) are more likely than the average American to be white, southern Baptist, and non-college educated, demographics that agree with Republicans on most issues (abortion, immigration, coal mining, welfare, etc.).
This is a very boring answer. You might be more interested in other, more interesting factors that contribute to reelection, which makes sense. But you need to keep in mind that this is the most important reason. The biggest takeaway from the past 100 years of polisci research is people support candidates with similar beliefs to them.
Because there aren't term limits. Combine this with the key points others have made -- low information voters, incumbent preference by the parties, etc. and you end up with the same idiots in office for decades.
It's funny -- people wonder why things never really change and then they vote for the same people over and over again.
How about a national movement -- if given the option in the primaries, don't vote for anyone who has ever held any public office! Either party... They are both as inbred and corrupt as they can get.
It might me a little chaotic initially, but I can't help but wonder if some meaningful things wouldn't get accomplished!
