Problems With Steemit's Built-in Search & How AskSteem Can Help

Introduction

About a month ago, an amazing Steem tool was introduced by @thekyle: Ask Steem - STEEM Search Engine. AskSteem is a unique tool that makes searching Steem's blockchain fairly easy.
There are just so many ways that you can query the blockchain. From simple queries that you normally use when searching on Google, like:
How to train your dog
to more complicated queries like:
((bitcoin OR tags:bitcoin) AND (net_votes:>500 AND children:(>50 AND <100))
For a more in-depth guide on how to use AskSteem, you can check the introduction post here: Introducing AskSteem - A steem search engine

Steemit's Built-in Search

If you've been on Steemit for a while, then more than likely you've used the Built-in Search tool more than once. To me personally, I was never impressed with this search tool as it normally brings old, usually irrelevant results. But why?

The current Steemit built-in search is powered by Google Custom Search. This is basically a tool that focuses on bringing the results for your queries from just 1 website, not the whole Google database. This is very similar to using the "site:" search operator:
site:steemit.com keywordgoeshere
However, using the Google Custom Search, you tend to get more results as fewer results are filtered out. IE: tag pages.

Now, let's do a quick comparison between using the "site:" operator on Google and using Steemit's Built-in Search for the same query steem power:

1- Using the site: operator:

Returned results: 118,000 results

2- Using Steemit's Built-in Search:

Returned results: 168,000 results

Now, comparing the two searches, you'll notice that the Built-in search brought in more results. It exactly returned 168,000 results while the "site:" operator only returned 118,000 results. That's 50,000 LESS returned results, which is HUGE!

So let's just stick with the Built-in Search tool, it works really well. Right? I don't think so!

But why? Well let me show you the results we should be expecting:

Returned results: 172,177

These are the total results for steem power in Steem's blockchain. Google Custom Search was close to the number of returned results but it failed in returning all of the results. It showed 4,177 fewer results and that's for just 1 keyword!

The problem with the Built-in Search tool is that it only brings the 'indexed' results in the Google database. For a page to be indexed, it requires the Google bot to visit that page, and it's up to Google to decide whether they want to add that page to their index or not. While I believe that Steemit.com holds a lot of authority and almost all pages are indexed within a few minutes of posting, but we're still relying on Google to make the decision.

So let's say a new post was made less than one minute ago, it's guaranteed to be found on AskSteem, but there's no guarantee at all that it'll ever be found on Google.

Another problem is the way that Google sorts the results. Let's check the first returned result for the same query steem power:

Now let's take a closer look at the post:

Now while I think it's a great post by @donkeypong, it's still a year old! So this doesn't exactly count as a 'fresh' result.

Another problem appears when we try to look for 'fresh' results. We can attempt to do so by sorting the results by date. Here's what we get when we try the sort by date option:

As you can see, the returned results have nothing to do with our query, NOT AT ALL! They were only returned as Google thinks they are fresh. And that's it.

The Solution

I believe we should all, as a community, start supporting AskSteem as I believe this is the way that search should be done. At the moment, AskSteem is no where near perfect as it still lacks a lot of things that would make the experience of searching a lot better.

For instance, the returned results by AskSteem are generally displayed in a random order, where in a perfect world (or a perfect search engine), some ranking factors should be introduced and used, like:

• Post date.
• Number of upvotes.
• Number of comments.
• Resteems.
• Author reputation / rank.
• Views.
• Social signals: facebook likes, twitter retweets, etc.
• Backlinks weight / number.
• On-site post mentions.

So I believe that an algorithm that takes into account all of these 'ranking factors' should be included in the development plans for AskSteem, as it will take the search experience to a whole another level. But this won't happen overnight and it won't happen without supporting the project.

---

That's it for today folks. And of course, if you like this post, please resetem, upvote and leave a comment.

Until another post, Steem on..

All the best,
@moataz