8

I'm about to write a paper that includes extensive searches on google with different queries. Is there any ways for me to show in my paper that this search provided these results at the time being?

3
  • 4
    Consider also asking this question in other SE communities, such as "Open Data" or "Cross Validated". Ultimately, this is a data technical question.
    – Scientist
    Commented Nov 14, 2018 at 11:19
  • 4
    Consider adding some more details. Google search results are hardly reproducible. They change over time, they change based on the language of the searching machine (even if the query is the same), based on the location of the IP address and possibly depending on the profile google has on the person.
    – Mefitico
    Commented Nov 14, 2018 at 15:18
  • 2
    How do other people in your field approach this issue?
    – Bryan Krause
    Commented Nov 14, 2018 at 18:59

2 Answers 2

8

Is there any ways for me to show in my paper that this [Google] search provided these results at the time being?

No, this isn't possible: the same search can simultaneously produce different results, because Google considers factors beyond the supplied search term, e.g., location.

That said, solutions are possible if you are willing to consider just the results of a single search. (Perhaps that's what you meant anyhow.) Leading to:

Yes (partly), you just need to record the raw result data (which will be authenticated, hence, cannot be forged*), which you can (probably) do using curl (with option --trace <filename>), for instance.

Ultimately, reproducibility is unattainable, but you can prove your results were derived by your method, as I have explained. (Your work might be useful without such proof.)


*There are some caveats.

5
  • 2
    user2768 raises a very important point. Google results are not identical for everybody. Not only do they vary geographically, they also depend on your search history and what other pages you visit on the internet, etc. If you want a search engine that does not use such information, duckduckgo.com is an alternative.
    – Phil
    Commented Nov 14, 2018 at 12:39
  • 3
    @Phil Indeed. I didn't mention personal search history, since this can be avoided, but location cannot (at least, a location cannot, the user's location can)
    – user2768
    Commented Nov 14, 2018 at 12:40
  • 2
    The same query executed at the same time on similar fresh-installs on the same hardware may produce different results if serviced by different Google data centers that each hold slightly different data at the time.
    – Jon Custer
    Commented Nov 14, 2018 at 14:33
  • 1
    I will describe the more I can about the conditions of my search, like time, date, city and etc. I just want to make a list of all the websites developed in a particular field in a particular city. And want to use search engines, directories, offline indications and etc. Commented Nov 15, 2018 at 5:47
  • 2
    @PooryaZeynalzadeh I edited to add a few details. Beyond those, you might want to consider building a website that periodically updates that list and write your paper about the website data at the time of writing.
    – user2768
    Commented Nov 15, 2018 at 7:12
2

I think copy and pasting the text or making screenshots is enough.

Google search results are out of your control, therefore not deterministic from your perspective.

If you want an additional integrity or certification, I think you are asking for too much. No research that I can think of comes with this kind of attestation. A biological research paper does not have an appendix with a sworn and signed statement saying: "I, cell 4588, hereby solemny swear that I divided 20% faster after being injected with chemical X."

Strictly speaking, nobody believes you until your results have been reproduced by other parties, anyway. Until then your paper is just a claim.

Seeing how Google search results are out of your control, the best you can do is describe what observations and deductions you made and then theorize on the hidden (deterministic?) model behind them. Screenshots or text results would then merely serve to illustrate your observations.

This is why having a detailed methodology section is important. The more details you provide, the better other scientist can approximate the circumstances of your experiments. Details could be your location, time, whether you cleared your cookies, whether you were signed in, operating system, etc.

2
  • This was in fact what I had in my mind, using screenshots and describing the exact details and info related to the conditions of my search. The goal is to find specific websites and pages developed for a specific city (which I'm searching from) and make a list of them. Commented Nov 15, 2018 at 5:44
  • 1
    Just my two cents: Manual copy-and-paste/screenshots don't scale, automated screenshots scale a bit (but limits will quickly be hit), automated copy-and-paste scales (albeit, it isn't really copy-and-paste, per se, at that point). That said, if you're automating, then why not authenticate too?
    – user2768
    Commented Nov 15, 2018 at 7:16

Not the answer you're looking for? Browse other questions tagged .