Citing wikipedia, and cherrypicking one of the few pieces of research that partially argues against the effect in the massive article, is dubious.

The study you cited, indeed, doesn't deny the bystander effect that each person becomes less likely to help as more people are present. It just argues that in the most dangerous situations, the odds of someone intervening goes up slightly as the number of people increase. See https://eprints.lancs.ac.uk/id/eprint/134891/1/Would_I_be_He...

I can tell you, my probability of intervening is lower the more people that are around. The total probability of someone intervening may go up or down based on a lot of circumstances. This is consistent with many individual research findings and meta-analyses, e.g. https://asset-pdf.scinapse.io/prod/2107050526/2107050526.pdf

As other people point out, it is rational and even helpful to be less inclined to intervene when more people are around and are likely to be able to help.

I cited wikipedia as that's where I learned that the "bystander effect" was a PR campaign invented by the NYPD in 1964, an interesting fact I thought to share. It states this in the first paragraph, then cites the study I chose to quote in both the next paragraph, as well as again later in the entry. I didn't cherry pick it, it's the main thrust of the page, and the largest study of its kind. In 219 observed disputes across 3 countries, the only times no bystander intervened were when couples were arguing amongst themselves, or when a thief was the one being beaten up.

It's hardly the main thrust of the page.

I linked the study above. Note that the individual probability of intervention falls as the number of people goes up (and the study you cited mentions this).

In some situations, it looks like the total probability of any intervention goes up as the number of people increases. In other situations, it looks like it goes down. The paper cited to produce that couple of sentences on Wikipedia makes the distinction quite clear and does not dispute the individual probability of intervention falls: https://eprints.lancs.ac.uk/id/eprint/134891/1/Would_I_be_He...

Note that it quite clearly says that it does not refute the bystander effect: ". It is important that by examining intervention on the situational rather than the individual level, our research does not evaluate whether bystanders are less likely to provide help when in the presence of other bystanders compared with when they are alone (i.e., the bystander effect). "

I agree and don't dispute the Kitty Genovese thing was bullshit, but the bystander effect-- a reduced probability for individuals to respond when more people are around-- is one of the most replicated findings in behavioral psychology.

As to the probability of someone intervening-- it looks like the odds are best when a small number of people are around vs. property crime, and best when a large number of people are around vs. dangerous/emergency situations. Also, the efficacy of the latter intervention, assessed by victims, seems to fall as the number of surrounding people increase.

This is all well discussed in the wikipedia article and in the text of the particular study you're relying upon.