Ever since Google demoted Scholar from the products bar, people have become anxious as to whether this is a sign that it will be discontinued. This feeling has only strengthened since the demise of Google Reader, which proved that Google is not afraid to discontinue services with (pretty) large and devoted audiences. Why would Google discontinue Scholar, and what would be the consequences?
Why killing Scholar might make sense
I see two reasons why discontinuing Google Scholar might make sense. First, consider Google is trying to focus on fewer products for more users (as evident from the closure of Reader); Scholar serves an academic niche. Indeed, many scholarly publications are already discoverable through the main Google Search. Other user groups in search of scholarly publications such as journalists prefer to use Google Search instead of Scholar . Scholars themselves are also heavy users of Search (see the figure below), and also find papers with it . Second, the academic landscape is moving from publishing to a wider range of online information. Researchers write and publish preprints, papers on their own webpages or social network profiles, datasets, programming code, software tools, lab notes, tweets and blogposts. The question of what is academic information is no longer trivial.
Disadvantages for searchers
The biggest disadvantage is the increasing burden upon researchers to decide which search results can be deemed of academic value and which not (a matter I previously raised in another blog post). Search engines already de-contextualize information ; academic information is not presented within its academic context, but within the context of the search environment. Replacing Scholar with Search would severely increase this de-contextualization, as publications are presented together with blogs, general websites, and so forth. However, this de-contextualization is already taking place as papers are decreasingly evaluated by the journal that published them . Moreover, with proposals such as altmetrics to complement the existing journal-based metrics or perhaps even replace them, one can wonder whether publications will remain a necessary or even sufficient context for academic information. Whether information found is valuable appears to become less dependent of its format, and more of its inherent value, making the aforementioned preprints, programming code, blog posts, etc. valuable academic assets.
Scholar versus Search
Will this lead to overcrowded search results, making it difficult to find publications? This is also true within Scholar, which can easily retrieve thousands of search results, even though Scholar’s precision matches that of specialized databases such as PubMed with scores around 8% (i.e. 8 out of 100 search results being relevant) . Thousands of publications is not necessarily easier to navigate than tens of thousands of mixed search results. In either case the searcher is dependent on the relevance ranking of the search algorithm  and has to assume that not all thousands of search results need to be considered. As such, doing an extensive and systematic literature search within Scholar is already ill-advised; although known-item searches work fairly well, discovery of all related papers should not be assumed . This is mainly due to the keyword-only search  employed by Scholar as well as Search, rendering it difficult to find related papers that employ terms different from the keywords in the query. In short, does Scholar solve a search scenario that cannot be done in Search?
Should online research be limited to publications, or will much broader content of academic material become available? If the latter is true, broadening searches to Search instead of Scholar might actually become preferable. Searching for publications will remain possible within Search, but will be complemented with the wealth of online academic output in other formats.
Cited by and related articles
Update 20-09-2013 13:30 One of the most mentioned features in the comments and on Twitter is the added value of “cited by” and “related articles” functionalities in Google Scholar. There is some confusion about whether this is available in Search or not. I’ve tried this in several Search domains (.com, .nl, .ca & .co.uk) with one of the papers referred below and got the same results every time: these functionalities are already available in Search. See this screenshot and notice the first three results. These features are powered by Google Scholar though; the question would be whether Search could be updated to incorporate these features natively.
Indexing deep web articles
Update 09-10-2013 12:00 One important difference between Search and Scholar is their treatment of deep web material. Where Search only indexes material that is directly accessible, Scholar does full-text indexing of journal articles that are behind pay-walls; a matter Robert Schonfeld raised in the comments. If this distinction is fundamental, this might answer my question about what search scenario solved by Scholar cannot be done in Search. However, what if Google changes their treatment of deep web material in Search? Since this blogpost, Robert Schonfeld has learned that Google might start indexing ScienceDirect (tweet1, tweet2). Thus, Google might actually already be taking the first steps to solve the issue of indexing journals in Search. This would appear to be a fundamental change for Search, but it might be a necessary step if Google wants to merge Scholar into Search.
@MaxJ_K Well I have learned that G Search is bringing in licensed resources, perhaps restricted to subscribers.
— Roger C. Schonfeld (@rschon) October 8, 2013
Update 18-11-2014: Google is not killing Scholar any time soon
To celebrate the tenth birthday of Google Scholar, Nature has done a wonderful interview with Anurag Acharya that addresses a lot of the points made in this blog post. To read the interview, click here. I have summarized the interview as a response in the following blog post: A Decade of Google Scholar.
 Kemman, M., Kleppe, M., Nieman, B., & Beunders, H. (2013). Dutch Journalism in the Digital Age. Icono 14, 11(2), 163–181.
 Kemman, M., Kleppe, M., & Scagliola, S. (2013). Just Google It – Digital Research Practices of Humanities Scholars. arXiv:1309.2434v1 [cs.DL]
 Jamali, H. R., & Asadi, S. (2010). Google and the scholar: the role of Google in scientists’ information-seeking behaviour. Online Information Review, 34(2), 282–294.
 Lozano, G. A., Lariviere, V., & Gingras, Y. (2012). The weakening relationship between the Impact Factor and papers’ citations in the digital age. arXiv:1205.4328v1 [cs.DL].
 Shariff, S. Z., Bejaimal, S. A., Sontrop, J. M., Iansavichus, A. V., Haynes, R. B., Weir, M. A., & Garg, A. X. (2013). Retrieving Clinical Evidence: A Comparison of PubMed and Google Scholar for Quick Clinical Searches. Journal of medical Internet research, 15(8).
 Pan, B., Hembrooke, H., Joachims, T., Lorigo, L., Gay, G., & Granka, L. (2007). In Google we trust: Users’ decisions on rank, position, and relevance. Journal of Computer‐Mediated Communication, 12(3), 801-823.
 Giustini, D., & Boulos, M. N. K. (2013). Google Scholar is not enough to be used alone for systematic reviews. Online journal of public health informatics,5(2), 214.
 Mann, T. (2008). Will Google’s Keyword Searching Eliminate the Need for LC Cataloging and Classification?. Journal of Library Metadata, 8(2), 159-168.
*Google Scholar was heavily used in the creation of this list of references