A next door neighbor in the 1940 census is listed with a wrong surname. Is it possible to have that corrected?
I assisted with a few pages of indexing for the 1940 census for a site other than ancestry. I can tell you that you can't change what's on the actual document created in the 1940's. You can suggest changes to the index for the census on sites like ancestry.com if you're a member. It will show up, after it's been approved as appropriate by the ancestry.com team. Indexing is often difficult since you're attempting to read someone's cursive handwriting and dealing with smudges and stains as well as misinformation on the document. Once it has been indexed, it is submitted to the team to be audited. If the auditor disagrees with the indexer's best guess, they will change it themselves.
Your suggestions to ancestry will not replace what is there, but will show up beneath it as an optional interpretation along with an explanation of how/why you feel it should be different. If you are an ancestry member, it helps if you add these interpretations of the material as they can help other people looking for the same family.
The 1940 census images we work with are historical records. As with all records and all historical records, the information contained therein is subject to error, oversight and omission.
Those of us engaged in preserving historical documents want them maintained in all their glory--misspellings, typos, tears, etc.; we wouldn't want to see "original content" altered.
See also JustinY's comment. If you are asking about whether the cataloging or indexing of a record can be improved or commented upon, that is a different question.
Whether it is possible to change the name that you believe is in error depends on where the information is held and by whom.
You obviously have an entry in your private collection that conflicts with other evidence that you hold or have seen. The first step is not to replace one or the other but to note the differences, to state where each piece of information came from, and to explain which one you believe to be correct and why.
The second level of action relates to the source of the information you reject. In this case, it has come from one of several indexes independently constructed for the 1940 Census. Do each of the other (free) indices give the name in the same way? Or is your source out of step with the others. Remember several groups used different people, different techniques and different quality assurance methods to complete the mammoth task of building their index. If the conflict comes from one source only, then you can go back to your notes and record that fact as strengthening the support for your preferred name (and perhaps reducing your trust in that source or repository). As a courtesy, you could advise the provider of the flawed information of how others read the record.
If all of the indexing projects have transcribed the name in the same way that does not match your other evidence, then you need to move on to examine the image of the original form. Can you see how several people could have made the same transcription error? If you are convinced that this is the case, you need to decide whether you want to advise everyone else they are wrong.
On the other hand, the "error" (if there is one) may have occurred at enumeration. If you have evidence that the family resident at that address at that time customarily used a different name or wrote the name in a different way, then continue to document that conflict.
At the end of all that, you can not change the name "in the census" because there is no name to change in what NARA holds. There is a series of marks made in 1940 to identify a person. The identifying information was separated from the other data about that person at an early stage in the processing to meet the requirements of confidentiality.
What we have is an image of an historical artefact that, as @GeneJ explained, is what it is. And is worthy of preservation unchanged.