Demanding a publication record in order to write a recommendation letter makes no sense at all to me.

Let us give an extreme example: suppose someone publishes in Cell by his/her third year in graduate school, and publishes again in Nature when he/she finishes the graduate program. This guy will not need a recommendation letter at all.

On the other hand, imagine a hard-working fellow who is knowledgeable, knows the field in depth and is good at designing experiments, but got unlucky because the biology of his/her hypothesis was not what he/she, nor the supervisor, were expecting and, because of that, did not publish in a high grade journal or did not publish at a fast pace. This fellow is the guy who needs someone to state his/her qualities and to be vouched for. The recommendation letter is redundant for people with high grade publications.

So, writing recommendations only for people with a good publication record is the same as pouring water into the sea.

Demanding publication record in order to write a recommendation letter has no point at all to me.

Let us give an extreme example: someone publishes in Cell by his/her third year in graduate school and publishes again in Nature when he/she finishes the graduate program. This guy will not need a recommendation letter at all.

On the other hand, imagine a hard-working fellow who is acknowledgeable, knows the field in depth and is good at designing experiments, but got unlucky because the biology of his/her hypothesis was not what he/she, neither the supervisor, were expecting and, because of that, did not publish in a high grade journal or did not publish in a fast pace. This fellow is the guy who needs someone to state his/her qualities and to be vouched for. The recommendation letter is redundant for people with high grade publications, whereas it is usually essential for people without them.

Demanding publication record in order to write a recommendation letter has no point at all to me.

Let us give an extreme example: someone publishes in Cell by his/her third year in graduate school and publishes again in Nature when he/she finishes the graduate program. This guy will not need a recommendation letter at all.

On the other hand, imagine a hard-working fellow who is acknowledgeable, knows the field in depth and is good at designing experiments, but got unlucky because the biology of his/her hypothesis was not what he/she, neither the supervisor, were expecting and, because of that, did not publish in a high grade journal or did not publish in a fast pace. This fellow is the guy who needs someone to state his/her qualities and to be vouched for. The recommendation letter is redundant for people with high grade publications.

So, writing recommendations only to people with a good publication record is the same as pouring water to the sea.

Demanding a publication record in order to write a recommendation letter makes no sense at all to me.

Let us give an extreme example: suppose someone publishes in Cell by his/her third year in graduate school, and publishes again in Nature when he/she finishes the graduate program. This guy will not need a recommendation letter at all.

On the other hand, imagine a hard-working fellow who is knowledgeable, knows the field in depth and is good at designing experiments, but got unlucky because the biology of his/her hypothesis was not what he/she, nor the supervisor, were expecting and, because of that, did not publish in a high grade journal or did not publish at a fast pace. This fellow is the guy who needs someone to state his/her qualities and to be vouched for. The recommendation letter is redundant for people with high grade publications.

So, writing recommendations only for people with a good publication record is the same as pouring water into the sea.