Bug in Search Results for Studies • page 2/2 • Lichess Feedback • lichess.org

@Rjw69 said in #8:

The search function in studies used to return relevant results.

I don't see how the recent changes can be represented as an improvement.

@Pawnstructures said in #9:

For me the recent changes feels more like a bug compared to the pre update experience.

The current search provide more relevant/better results in general (minus chapter texts), so in study search (https://lichess.org/study/search) the results are much less random then before.

But as you both said, it breaks some existing user experiences without providing an alternative, which is very bad. I Apologize this. As my post above, this will be improved. Please be patient!

@Rjw69 said in #8: > The search function in studies used to return relevant results. > > I don't see how the recent changes can be represented as an improvement. @Pawnstructures said in #9: > For me the recent changes feels more like a bug compared to the pre update experience. The current search provide more relevant/better results in general (minus chapter texts), so in study search (https://lichess.org/study/search) the results are much less random then before. But as you both said, it breaks some existing user experiences without providing an alternative, which is very bad. I Apologize this. As my post above, this will be improved. Please be patient!

CM Pawnstructures edited

#12

Hi Thanh! I couldnt DM you because your inbox was not accepting messages, I want to explain exactly how I use study search in my day-to-day coaching, and why the recent change breaks an important part of my workflow. I’m a coach and I currently have around 1000 studies and multiple students each week. This means I need a reliable way to check whether I have already shown a specific game or chapter to a particular student. Here is a real example from this week with my student Mr X. A couple of months ago I showed him an unusual Sicilian line. Before preparing yesterday’s lesson, I wanted to add a chapter with a game where I won against a GM when i played this unusual line. To avoid duplicating material, I always double-check whether the game has already appeared in any of the studies I have used with that student, he has been a student for a year so searching for studies where he is added as a contributor would return a lot of hits. My method is simple:

I search across my studies for the opponent’s username in this case “Ucitelot”, which is extremely uncommon.

My search looked like this:
owner:Pawnstructures Ucitelot

Under the old search system, this would return hits even if the username only appeared inside a chapter, not in the headline.

This was perfect, because the presence or absence of that username told me immediately whether the game had been shown to that student before.With the new change, the results came back empty, so I assumed the game had never been included. I added it to Mr X study.

When we reached that chapter during the lesson, he said the name looked familiar and after 10 moves he recognized the exact position and even remembered a missed tactic (...Nc6 being loose, Ng5, sacrifice on e6 etc.). So I had accidentally shown the same game twice, which is embarrasing and undermines the quality control I try to maintain. Trying to keep it short :) Regards Richard

Hi Thanh! I couldnt DM you because your inbox was not accepting messages, I want to explain exactly how I use study search in my day-to-day coaching, and why the recent change breaks an important part of my workflow. I’m a coach and I currently have around 1000 studies and multiple students each week. This means I need a reliable way to check whether I have already shown a specific game or chapter to a particular student. Here is a real example from this week with my student Mr X. A couple of months ago I showed him an unusual Sicilian line. Before preparing yesterday’s lesson, I wanted to add a chapter with a game where I won against a GM when i played this unusual line. To avoid duplicating material, I always double-check whether the game has already appeared in any of the studies I have used with that student, he has been a student for a year so searching for studies where he is added as a contributor would return a lot of hits. My method is simple: I search across my studies for the opponent’s username in this case “Ucitelot”, which is extremely uncommon. My search looked like this: owner:Pawnstructures Ucitelot Under the old search system, this would return hits even if the username only appeared inside a chapter, not in the headline. This was perfect, because the presence or absence of that username told me immediately whether the game had been shown to that student before.With the new change, the results came back empty, so I assumed the game had never been included. I added it to Mr X study. When we reached that chapter during the lesson, he said the name looked familiar and after 10 moves he recognized the exact position and even remembered a missed tactic (...Nc6 being loose, Ng5, sacrifice on e6 etc.). So I had accidentally shown the same game twice, which is embarrasing and undermines the quality control I try to maintain. Trying to keep it short :) Regards Richard

nt9

#13

thanks for the details response Richard, I'll prioritize to fix this issue, but unfortunately can't promise when it is fixed.

But I have one more question, when you search for chapter texts, do you expect them in the name, tags or also something like comments in the chapter's pgn?

@Pawnstructures said in #12:

I couldnt DM you because your inbox was not accepting messages,

oops, sorry I forgot I have that settings

thanks for the details response Richard, I'll prioritize to fix this issue, but unfortunately can't promise when it is fixed. But I have one more question, when you search for chapter texts, do you expect them in the name, tags or also something like comments in the chapter's pgn? @Pawnstructures said in #12: > I couldnt DM you because your inbox was not accepting messages, oops, sorry I forgot I have that settings

NM petaka

#14

I'm also having trouble with the changes to the search function in the studies. I was using it the same way as @Pawnstructures.

nt9

#15

@petaka said in #14:

I'm also having trouble with the changes to the search function in the studies. I was using it the same way as

Sorry to hear that, I'm working on a fix, will take one or two weeks. Hope that you can use it with sorting to mitigate this issue temporarily.

Note that, study search will only search for: name, description, topics, owner, member.

@petaka said in #14: > I'm also having trouble with the changes to the search function in the studies. I was using it the same way as Sorry to hear that, I'm working on a fix, will take one or two weeks. Hope that you can use it with sorting to mitigate this issue temporarily. Note that, study search will only search for: `name`, `description`, `topics`, `owner`, `member`.

@Rjw69 said in #16:

A search of any "subject" under relevant studies now returns a string of useless random studies that all have 1 heart and were written 3 yrs ago. They tend to be light on content, but all have the "subject" you are looking for in the title.

Yes, but you can sort it, by popular and then you have more interesting studies. btw what are you looking for when you're searching for subject? is this your own studies or you're a member of?

Dropping the information from inside the studies is a mistake, it makes it more difficult for users to find relevant subject matter and negatively impacts both study creators as well as the users of those studies.

It depends, before for example, if I searched for french defense, A lot of things showed up randomly like some Karo Can defense study because there is some comment somewhere says that karo can is better than french. Which makes sorting is totally useless because top studies usually contains a lot of text.

Anyone who actually used the searches that were previously available to look for content will know this.

Yes, I agreed this breaks people's workflow and habit.

What I can't understand is if some new search function for studies is needed, and it probably is, why does that mean that the other search functions that were already working had to be destroyed.

As I answered before, searching chapter texts means sorting useless.

@Rjw69 said in #16: > A search of any "subject" under relevant studies now returns a string of useless random studies that all have 1 heart and were written 3 yrs ago. They tend to be light on content, but all have the "subject" you are looking for in the title. Yes, but you can sort it, by popular and then you have more interesting studies. btw what are you looking for when you're searching for subject? is this your own studies or you're a member of? > Dropping the information from inside the studies is a mistake, it makes it more difficult for users to find relevant subject matter and negatively impacts both study creators as well as the users of those studies. It depends, before for example, if I searched for french defense, A lot of things showed up randomly like some `Karo Can defense` study because there is some comment somewhere says that karo can is better than french. Which makes sorting is totally useless because top studies usually contains a lot of text. > Anyone who actually used the searches that were previously available to look for content will know this. Yes, I agreed this breaks people's workflow and habit. > What I can't understand is if some new search function for studies is needed, and it probably is, why does that mean that the other search functions that were already working had to be destroyed. As I answered before, searching chapter texts means sorting useless.

bughouse26

#18

In the short term can we revert the change which broke content search and re-introduce the sorting feature once that doesn't introduce a regression?

As I answered before, searching chapter texts means sorting useless.

I typically search studies for chapter text content; I don't care about sort order so would prefer that the old behavior be restored. Sorting would be a nice-to-have in addition once it doesn't break existing search functionality.

In the short term can we revert the change which broke content search and re-introduce the sorting feature once that doesn't introduce a regression? > As I answered before, searching chapter texts means sorting useless. I typically search studies for chapter text content; I don't care about sort order so would prefer that the old behavior be restored. Sorting would be a nice-to-have in addition once it doesn't break existing search functionality.