> For the complete documentation index, see [llms.txt](https://nextsearch.gitbook.io/next-proposal/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://nextsearch.gitbook.io/next-proposal/whats-wrong-with-web-search/poor-quality-of-search-results.md).

# Poor Quality of Search Results

The results returned by search engines have improved significantly over prior engines BUT are still **abysmal** in so many cases. Disagree? Here are a few simple queries you can make to see the poor quality of current search results:

## Query: History of Search Engines

**Query**: *History of Search Engines*

**Results**: Notice the results that are outdated, poorly presented, have duplicate content, or are covering SEO rather than search engines properly.

## Query: "Best {Product/Service}"

**Query**: *Best {Product/Service}*

* Choose whatever product or service you like - refrigerator, laptop, bed, headphones, etc.

**Results**:

* Most reviews come from "big entity" sites - the indie/personal web is missing.
* It is easy to end up reading the same/similar review on multiple sites before finding unique/helpful content.
* Some sites are no-name brands that may have questionable practices (e.g., not actually testing products that are reviewed, reviews that drive the viewer to the highest commission generating product).
* Oftentimes the length of the articles and the details shared are of interest to search engines but unnecessarily verbose for humans.

## Query: "CNN" or "Fox News" or "{Organization Name}"

**Query**: *CNN* or *Fox News* or *{Organization Name}*

**Results**:

* Often dominated by the organization itself (not entirely unreasonably).
* Many organizations (e.g. news) have duplicate content from sites syndicating their content. Do we really want to see the same story from CNN, APNews, and MSNBC?
* One can use query operators to refine the results (but how many people do you know who use them or are even aware that they exist?). 
* Even if one uses query operators the number of sites that need to be excluded can become quite lengthy and must be recreated for each query.

## Query: {Medical Topic}

**Query**: {Any Medical Topic}

**Results**: Medical topics, oftentimes of critical importance to the searcher, return notoriously poor results. These results suffer from information that is:

* Incorrect - Oftentimes sincere, this information is untested and sometimes dangerous.
* Incomplete - Does not provide enough information for an individual to take informed action.
* Redundant - A number of larger medical sites provide primarily redundant information regarding the cause, diagnosis, treatment, prospects, etc.
* Commercial Information - Treatments of questionable value are marketed to individuals at high costs.

## Conclusion

We could go on *ad nauseum* with examples of poor quality search results but we don't want to bore you.

If you need more examples, [reach out to us](https://nextsearch.gitbook.io/introduction/request-for-comments.md).

## For Further Consideration

* Keselman A, Arnott Smith C, Murcko AC, Kaufman DR. [Evaluating the Quality of Health Information in a Changing Digital Ecosystem](https://www.jmir.org/2019/2/e11129). Journal of Medical Internet Research 2019;21(2):e11129. DOI: 10.2196/11129.
* Jimenez G, Lum E, Car J. [Examining Diabetes Management Apps Recommended From a Google Search: Content Analysis](https://mhealth.jmir.org/2019/1/e11848). JMIR mHealth and uHealth 2019;7(1):e11848. DOI: 10.2196/11848.
* Le H, Maragh R, Ekdale B, High A, Havens T, and Shafiq Z. [Measuring Political Personalization of Google News Search. In The World Wide Web Conference (WWW '19)](https://doi.org/10.1145/3308558.3313682). Association for Computing Machinery, 2019; 2597-2963. DOI: 10.1145/3308558.3313682.
* Metaxa D, Park JS, Landay JA, and Hancock J. [Search Media and Elections: A Longitudinal Investigation of Political Search Results](https://doi.org/10.1145/3359231). Proceedings of the ACM Human-Computer Interaction 3, CSCW, Article 129 (November 2019), 17 pages. DOI: 10.1145/3359231.