gistfile1.js

const Random = require('random-js')

const random = new Random(Random.engines.mt19937().autoSeed())

const values = ['2', '3', '4', '5', '6', '7', '8', '9', 'T', 'J', 'Q', 'K', 'A']
const suits = ['d', 'h', 'c', 's']

let originalDeck =
  suits.reduce((memo, suit) =>
    [...memo, ...values.map(value => `${value}${suit}`)], []
  )

// EXAMPLE OF THE MINIMAL STATE
// {
//   players: {
//     <id> : { status: 'inGame' }
//   },
//   seats: [[<position>, <id>], [<position>, <id>],
//   playersCards: {}
// }
function shuffleAndGiveCards (state) {
  let deck = random.shuffle([...originalDeck])

  originalDeck = [...deck]

  let playersCards = {}

  let inGamePlayersSeats =
    state
      .seats
      .reduce((memo, seat) => {
        if (state.players[seat[1]].status === 'inGame') {
          memo.push(seat[1])
        }

        return memo
      }, [])

  inGamePlayersSeats
    .forEach((playerID, playerIndex) => {
      let cardsIndex = [
        playerIndex,
        (playerIndex + inGamePlayersSeats.length)
      ]

      playersCards[playerID] = {
        cards: [
          deck[cardsIndex[0]],
          deck[cardsIndex[1]]
        ],
        hiding: true
      }

      deck[cardsIndex[0]] = null
      deck[cardsIndex[1]] = null
    })

  deck = deck.filter(card => card)

  return {
    ...state,
    playersCards,
    deck
  }
}

Are you concerned about the underlying implementation of autoSeed here? Millisecond time stamp + the output of Math.random() doesn't seem like a great seeding source. I don't know if your runtime supports crypto.getRandomValues, but it might be a better choice for seeding.

I.e., something like this would better seed mersenne twister, assuming crypto has a high quality source:

const Crypto = require('crypto')

const seed = new Uint32Array(624);
let dataView = new DataView(seed.buffer);
Crypto.randomFillSync(dataView, dataView.byteOffset, dataView.byteLength)

const random = new Random(Random.engines.mt19937().seedWithArray(seed))

This should only be done once per instance, as it is probably fairly slow.

What was the result? I agree with your assessment and wonder if the implementation of the autoseed has changed