Completion suggester returning incorrect scores

I am playing around with the completion suggester and I am not getting the scores i expect from it

To reproduce this

  1. create the index

    PUT mytest

    PUT mytest/suggestion/_mapping
    {
    "suggestion" : {
    "_source": {
    "enabled": false
    },
    "_all": {
    "enabled": false
    },
    "properties" : {
    "keyword" : {
    "type" : "completion",
    "analyzer" : "keyword",
    "context": {
    "category_id": {
    "type": "category",
    "default": 0
    }
    }
    }
    }
    }
    }

  2. add 2 documents to the index

    PUT mytest/suggestion/1?refresh=true
    {
    "keyword" : {
    "input": "mercedes",
    "context": {
    "category_id": 91
    },
    "weight" : 53477373
    }
    }

    PUT mytest/suggestion/2?refresh=true
    {
    "keyword" : {
    "input": "mazda",
    "context": {
    "category_id": 91
    },
    "weight" : 53477371
    }
    }

  3. Query the for the documents

    POST mytest/_suggest
    {
    "autocomplete-suggest" : {
    "text" : "m",
    "completion" : {
    "field" : "keyword",
    "context": {
    "category_id": 91
    }
    }
    }
    }

This will give the following result

{
   "_shards": {
      "total": 1,
      "successful": 1,
      "failed": 0
   },
   "autocomplete-suggest": [
      {
         "text": "m",
         "offset": 0,
         "length": 1,
         "options": [
            {
               "text": "mazda",
               "score": 53477372
            },
            {
               "text": "mercedes",
               "score": 53477372
            }
         ]
      }
   ]
}

Clearly for mazda the score should be 53477371 and for mercedes the score should be 53477373 but in both cases elastic decides that the score is 53477372

Can somebody explain why this is happening ?

I am using elastic search 1.7.1