(3B) Nonscaling Delay Curve.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "92bffbfd-e89b-45a2-9174-0fd203e96d36",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loaded\n"
     ]
    }
   ],
   "source": [
    "import dill\n",
    "import math\n",
    "import numpy as np\n",
    "from pprint import pprint\n",
    "from bayes_opt import BayesianOptimization\n",
    "from datetime import datetime as dt\n",
    "\n",
    "delays = {}\n",
    "\n",
    "exec(open('utils.py').read())\n",
    "print('Loaded')\n",
    "\n",
    "MAX_YEAR = CURRENT_YEAR + 100 # CURRENT_YEAR defined in utils.py\n",
    "years = list(range(CURRENT_YEAR, MAX_YEAR))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fb92a6f5-c3a8-40be-89ae-0b29c39d3fc0",
   "metadata": {},
   "source": [
    "## Algorithms/insights delay"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "ede8a8fc-b4a4-4273-9b44-913170dda139",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "|   iter    |  target   |   push    |   shift   |   slope   |\n",
      "-------------------------------------------------------------\n",
      "| \u001b[95m6        \u001b[0m | \u001b[95m-0.03867 \u001b[0m | \u001b[95m0.6241   \u001b[0m | \u001b[95m7.728    \u001b[0m | \u001b[95m8.543    \u001b[0m |\n",
      "| \u001b[95m41       \u001b[0m | \u001b[95m-0.03702 \u001b[0m | \u001b[95m0.01     \u001b[0m | \u001b[95m7.907    \u001b[0m | \u001b[95m5.513    \u001b[0m |\n",
      "| \u001b[95m47       \u001b[0m | \u001b[95m-0.03702 \u001b[0m | \u001b[95m0.01     \u001b[0m | \u001b[95m10.0     \u001b[0m | \u001b[95m5.63     \u001b[0m |\n",
      "| \u001b[95m55       \u001b[0m | \u001b[95m-0.03686 \u001b[0m | \u001b[95m0.01     \u001b[0m | \u001b[95m5.845    \u001b[0m | \u001b[95m5.217    \u001b[0m |\n",
      "| \u001b[95m56       \u001b[0m | \u001b[95m-0.03404 \u001b[0m | \u001b[95m0.01     \u001b[0m | \u001b[95m2.914    \u001b[0m | \u001b[95m4.004    \u001b[0m |\n",
      "| \u001b[95m58       \u001b[0m | \u001b[95m-0.01643 \u001b[0m | \u001b[95m0.01     \u001b[0m | \u001b[95m3.706    \u001b[0m | \u001b[95m2.759    \u001b[0m |\n",
      "| \u001b[95m69       \u001b[0m | \u001b[95m-0.01337 \u001b[0m | \u001b[95m0.02311  \u001b[0m | \u001b[95m7.292    \u001b[0m | \u001b[95m2.791    \u001b[0m |\n",
      "=============================================================\n",
      "Curve params found\n",
      "{'push': 0.02310948127282846,\n",
      " 'shift': 7.291764982890799,\n",
      " 'slope': 2.7911121542768265}\n",
      "-\n",
      "## Chance of nonscaling delay ##\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Year: 2025 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 20%\n",
      "Year: 2026 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 20%\n",
      "Year: 2027 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 15%\n",
      "Year: 2028 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 6%\n",
      "Year: 2029 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2030 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2031 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2032 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2033 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2034 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2035 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2045 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2055 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2065 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2075 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2085 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2095 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2105 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2115 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "-\n",
      "{1: 1,\n",
      " 5: 1,\n",
      " 10: 1,\n",
      " 20: 2,\n",
      " 30: 2,\n",
      " 40: 2,\n",
      " 50: 2,\n",
      " 60: 3,\n",
      " 70: 3,\n",
      " 80: 4,\n",
      " 90: 5,\n",
      " 95: 7,\n",
      " 99: 16}\n"
     ]
    }
   ],
   "source": [
    "nonscaling_points = [(2025, 0.2),\n",
    "                     (2027, 0.15),\n",
    "                     (2030, 0.1),\n",
    "                     (2050, 0.05)]\n",
    "p_nonscaling_delay = derive_nonscaling_delay_curve(nonscaling_points)\n",
    "delays['algorithms'] = {'prob': p_nonscaling_delay, 'length': sq.mixture([[0.9, sq.lognorm(1, 5)], [0.1, sq.lognorm(2, 20)]])}\n",
    "plot_nonscaling_delay(plt, years, p_nonscaling_delay)\n",
    "print('-')\n",
    "pprint(sq.get_percentiles(delays['algorithms']['length'] @ 1000, digits=0))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "75f3fbca-072c-489c-a1d6-6cfbf43b569c",
   "metadata": {},
   "source": [
    "## Infrastructure delay"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "027d2084-28bb-4f50-9cd9-4a30fd661a09",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "|   iter    |  target   |   push    |   shift   |   slope   |\n",
      "-------------------------------------------------------------\n",
      "| \u001b[95m2        \u001b[0m | \u001b[95m-0.3085  \u001b[0m | \u001b[95m1.841    \u001b[0m | \u001b[95m1.548    \u001b[0m | \u001b[95m7.556    \u001b[0m |\n",
      "| \u001b[95m3        \u001b[0m | \u001b[95m-0.2454  \u001b[0m | \u001b[95m1.275    \u001b[0m | \u001b[95m9.695    \u001b[0m | \u001b[95m9.981    \u001b[0m |\n",
      "| \u001b[95m8        \u001b[0m | \u001b[95m-0.2131  \u001b[0m | \u001b[95m0.4449   \u001b[0m | \u001b[95m4.901    \u001b[0m | \u001b[95m4.742    \u001b[0m |\n",
      "| \u001b[95m26       \u001b[0m | \u001b[95m-0.1324  \u001b[0m | \u001b[95m0.7256   \u001b[0m | \u001b[95m9.41     \u001b[0m | \u001b[95m2.397    \u001b[0m |\n",
      "| \u001b[95m29       \u001b[0m | \u001b[95m-0.1199  \u001b[0m | \u001b[95m1.222    \u001b[0m | \u001b[95m8.66     \u001b[0m | \u001b[95m0.6169   \u001b[0m |\n",
      "| \u001b[95m33       \u001b[0m | \u001b[95m-0.07854 \u001b[0m | \u001b[95m0.3404   \u001b[0m | \u001b[95m2.136    \u001b[0m | \u001b[95m1.449    \u001b[0m |\n",
      "| \u001b[95m43       \u001b[0m | \u001b[95m-0.04347 \u001b[0m | \u001b[95m1.035    \u001b[0m | \u001b[95m9.887    \u001b[0m | \u001b[95m1.041    \u001b[0m |\n",
      "| \u001b[95m66       \u001b[0m | \u001b[95m-0.02973 \u001b[0m | \u001b[95m0.5471   \u001b[0m | \u001b[95m9.175    \u001b[0m | \u001b[95m1.448    \u001b[0m |\n",
      "=============================================================\n",
      "Curve params found\n",
      "{'push': 0.5470753332396583,\n",
      " 'shift': 9.17549854793724,\n",
      " 'slope': 1.4479367285117843}\n",
      "-\n",
      "## Chance of nonscaling delay ##\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Year: 2025 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 90%\n",
      "Year: 2026 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 80%\n",
      "Year: 2027 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 50%\n",
      "Year: 2028 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 21%\n",
      "Year: 2029 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 9%\n",
      "Year: 2030 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 6%\n",
      "Year: 2031 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2032 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2033 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2034 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2035 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2045 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2055 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2065 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2075 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2085 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2095 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2105 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2115 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "-\n",
      "{1: 1,\n",
      " 5: 1,\n",
      " 10: 1,\n",
      " 20: 1,\n",
      " 30: 2,\n",
      " 40: 2,\n",
      " 50: 2,\n",
      " 60: 3,\n",
      " 70: 3,\n",
      " 80: 3,\n",
      " 90: 4,\n",
      " 95: 5,\n",
      " 99: 7}\n"
     ]
    }
   ],
   "source": [
    "# https://twitter.com/ohlennart/status/1645058017119854592\n",
    "# https://asteriskmag.com/issues/03/the-transistor-cliff\n",
    "nonscaling_points = [(2025, 0.90),  # 0.2 power, 0.15 GPU supply, 0.2 latency/distributed, 0.8 buildup, 0.05 other => 1 - ((1 - 0.2) * (1 - 0.15) * (1 - 0.2) * (1 - 0.8) * (1 - 0.05)) = 0.90\n",
    "                     (2026, 0.74), # 0.15 power, 0.1 GPU supply, 0.1 latency/distributed, 0.6 buildup, 0.05 other =>  1 - ((1 - 0.15) * (1 - 0.1) * (1 - 0.1) * (1 - 0.6) * (1 - 0.05)) = 0.74\n",
    "                     (2027, 0.46), # 0.1 power, 0.05 GPU supply, 0.05 latency/distributed, 0.3 buildup, 0.05 other => 1 - ((1 - 0.1) * (1 - 0.05) * (1 - 0.05) * (1 - 0.3) * (1 - 0.05)) = 0.46\n",
    "                     (2028, 0.23), # 0.05 power, 0.02 GPU supply, 0.05 latency/distributed, 0.1 buildup, 0.03 other =>  1 - ((1 - 0.05) * (1 - 0.02) * (1 - 0.05) * (1 - 0.1) * (1 - 0.03)) = 0.23\n",
    "                     (2030, 0.11),  # 0.02 power, 0.01 GPU supply, 0.02 latency/distributed, 0.05 buildup, 0.02 other => 1 - ((1 - 0.02) * (1 - 0.01) * (1 - 0.02) * (1 - 0.05) * (1 - 0.02)) = 0.11\n",
    "                     (2035, 0.05)] # 0.01 power, 0.01 GPU supply, 0.01 latency/distributed, 0.01 buildup, 0.01 other => 1 - ((1 - 0.01) * (1 - 0.01) * (1 - 0.01) * (1 - 0.01) * (1 - 0.01)) = 0.05\n",
    "\n",
    "p_nonscaling_delay = derive_nonscaling_delay_curve(nonscaling_points)\n",
    "delays['infra'] = {'prob': p_nonscaling_delay, 'length': sq.lognorm(1, 5)}\n",
    "plot_nonscaling_delay(plt, years, p_nonscaling_delay)\n",
    "print('-')\n",
    "pprint(sq.get_percentiles(delays['infra']['length'] @ 1000, digits=0))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ff1615c5-bb32-487f-a750-dbc41fb8d00e",
   "metadata": {},
   "source": [
    "## Data delay"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "568e938a-922d-43fc-8382-6bca86332e3c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "|   iter    |  target   |   push    |   shift   |   slope   |\n",
      "-------------------------------------------------------------\n",
      "| \u001b[95m3        \u001b[0m | \u001b[95m-0.009649\u001b[0m | \u001b[95m5.001    \u001b[0m | \u001b[95m8.083    \u001b[0m | \u001b[95m9.133    \u001b[0m |\n",
      "| \u001b[95m8        \u001b[0m | \u001b[95m-0.009627\u001b[0m | \u001b[95m3.654    \u001b[0m | \u001b[95m4.054    \u001b[0m | \u001b[95m7.426    \u001b[0m |\n",
      "| \u001b[95m13       \u001b[0m | \u001b[95m-0.003795\u001b[0m | \u001b[95m1.404    \u001b[0m | \u001b[95m5.887    \u001b[0m | \u001b[95m6.571    \u001b[0m |\n",
      "| \u001b[95m19       \u001b[0m | \u001b[95m-0.002078\u001b[0m | \u001b[95m1.202    \u001b[0m | \u001b[95m9.754    \u001b[0m | \u001b[95m6.692    \u001b[0m |\n",
      "| \u001b[95m26       \u001b[0m | \u001b[95m-0.000330\u001b[0m | \u001b[95m0.3857   \u001b[0m | \u001b[95m3.357    \u001b[0m | \u001b[95m7.606    \u001b[0m |\n",
      "| \u001b[95m38       \u001b[0m | \u001b[95m-1.152e-0\u001b[0m | \u001b[95m0.2285   \u001b[0m | \u001b[95m7.712    \u001b[0m | \u001b[95m9.78     \u001b[0m |\n",
      "| \u001b[95m41       \u001b[0m | \u001b[95m-0.0     \u001b[0m | \u001b[95m0.01     \u001b[0m | \u001b[95m5.002    \u001b[0m | \u001b[95m9.761    \u001b[0m |\n",
      "=============================================================\n",
      "Curve params found\n",
      "{'push': 0.01, 'shift': 5.001623822994236, 'slope': 9.761099247711758}\n",
      "-\n",
      "## Chance of nonscaling delay ##\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Year: 2025 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 5%\n",
      "Year: 2026 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 2%\n",
      "Year: 2027 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 2%\n",
      "Year: 2028 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 2%\n",
      "Year: 2029 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 2%\n",
      "Year: 2030 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 2%\n",
      "Year: 2031 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 2%\n",
      "Year: 2032 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 2%\n",
      "Year: 2033 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 2%\n",
      "Year: 2034 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 2%\n",
      "Year: 2035 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 2%\n",
      "Year: 2045 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 2%\n",
      "Year: 2055 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 2%\n",
      "Year: 2065 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 2%\n",
      "Year: 2075 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 2%\n",
      "Year: 2085 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 2%\n",
      "Year: 2095 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 2%\n",
      "Year: 2105 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 2%\n",
      "Year: 2115 - chance of a nonscaling delay if TAI compute needs are otherwise met in this year: 2%\n",
      "-\n",
      "{1: 1,\n",
      " 5: 1,\n",
      " 10: 1,\n",
      " 20: 1,\n",
      " 30: 2,\n",
      " 40: 2,\n",
      " 50: 2,\n",
      " 60: 3,\n",
      " 70: 3,\n",
      " 80: 3,\n",
      " 90: 4,\n",
      " 95: 5,\n",
      " 99: 7}\n"
     ]
    }
   ],
   "source": [
    "nonscaling_points = [(2025, 0.05),\n",
    "                     (2030, 0.02)]\n",
    "p_nonscaling_delay = derive_nonscaling_delay_curve(nonscaling_points)\n",
    "delays['data'] = {'prob': p_nonscaling_delay, 'length': sq.lognorm(1, 5)}\n",
    "plot_nonscaling_delay(plt, years, p_nonscaling_delay)\n",
    "print('-')\n",
    "pprint(sq.get_percentiles(delays['data']['length'] @ 1000, digits=0))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b124feaa-b239-4982-acbb-d3efc6054bcc",
   "metadata": {},
   "source": [
    "## Robotics need delay"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "3b74ba6e-ff8b-4311-bd16-fbdaaf881bd3",
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\"\n",
    "nonscaling_points = [(2024, 0.3),\n",
    "                     (2030, 0.2),\n",
    "                     (2040, 0.1),\n",
    "                     (2050, 0.05)]\n",
    "p_nonscaling_delay = derive_nonscaling_delay_curve(nonscaling_points)\n",
    "delays['robotics'] = {'prob': p_nonscaling_delay, 'length': sq.mixture([[0.85, sq.lognorm(1, 5)],\n",
    "                                                                        [0.15, sq.lognorm(1, 15)]])}\n",
    "plot_nonscaling_delay(plt, years, p_nonscaling_delay)\n",
    "print('-')\n",
    "pprint(sq.get_percentiles(delays['robotics']['length'] @ 1000, digits=0))\n",
    "\"\"\"\n",
    "None"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "962df8a5-7d24-4001-96d0-1b8ec1a0f741",
   "metadata": {},
   "source": [
    "## Integration delay"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "ec560708-06f9-4b0c-8af7-04f2785650df",
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\"\n",
    "# https://twitter.com/MatthewJBar/status/1653461856919654400\n",
    "nonscaling_points = [(2024, 0.4),\n",
    "                     (2030, 0.2),\n",
    "                     (2040, 0.1),\n",
    "                     (2050, 0.05)]\n",
    "p_nonscaling_delay = derive_nonscaling_delay_curve(nonscaling_points)\n",
    "delays['integration'] = {'prob': p_nonscaling_delay, 'length': sq.lognorm(3, 30)}\n",
    "plot_nonscaling_delay(plt, years, p_nonscaling_delay)\n",
    "print('-')\n",
    "pprint(sq.get_percentiles(delays['integration']['length'] @ 1000, digits=0))\n",
    "\"\"\"\n",
    "None"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bfa3c990-2427-45bd-bfc5-0cf2917095aa",
   "metadata": {},
   "source": [
    "##  Real world feedback delay"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "e3010b54-63f4-46d5-8188-d51ec573ae45",
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\"\n",
    "nonscaling_points = [(2024, 0.1),\n",
    "                     (2030, 0.05),\n",
    "                     (2050, 0.02)]\n",
    "p_nonscaling_delay = derive_nonscaling_delay_curve(nonscaling_points)\n",
    "delays['real_world_feedback'] = {'prob': p_nonscaling_delay, 'length': sq.mixture([[0.8, sq.lognorm(3, 10)],\n",
    "                                                                                   [0.1, sq.lognorm(10, 30)],\n",
    "                                                                                   [0.1, sq.lognorm(20, 50)]])}\n",
    "plot_nonscaling_delay(plt, years, p_nonscaling_delay)\n",
    "print('-')\n",
    "pprint(sq.get_percentiles(delays['real_world_feedback']['length'] @ 1000, digits=0))\n",
    "\"\"\"\n",
    "None"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b100c5f5-5c16-4515-8814-ae81391126d9",
   "metadata": {},
   "source": [
    "## Take off speeds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "acad56b4-027e-456e-8569-4a38812ee5a3",
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\"\n",
    "# https://www.lesswrong.com/posts/Gc9FGtdXhK9sCSEYu/what-a-compute-centric-framework-says-about-ai-takeoff\n",
    "nonscaling_points = [(2024, 0.7),\n",
    "                     (2030, 0.5),\n",
    "                     (2050, 0.35),\n",
    "                     (2080, 0.2)]\n",
    "p_nonscaling_delay = derive_nonscaling_delay_curve(nonscaling_points)\n",
    "delays['take_off'] = {'prob': p_nonscaling_delay, 'length': sq.lognorm(2, 15)}\n",
    "plot_nonscaling_delay(plt, years, p_nonscaling_delay)\n",
    "print('-')\n",
    "pprint(sq.get_percentiles(delays['take_off']['length'] @ 1000, digits=0))\n",
    "\"\"\"\n",
    "None"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a6bb37d3-59fd-48ab-b34d-634163fdd778",
   "metadata": {},
   "source": [
    "## Save"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "3a5ee983-96e5-4392-a1b4-44588eba845e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cached!\n"
     ]
    }
   ],
   "source": [
    "with open('caches/nonscaling_delays.dill', 'wb') as f:\n",
    "    dill.dump(delays, f)\n",
    "    \n",
    "print('cached!')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "911cabe6-7f04-4bc3-98ac-7c6702c6ac94",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "File last ran: 2025-01-23 11:52:18.921842\n"
     ]
    }
   ],
   "source": [
    "print('File last ran: {}'.format(dt.now()))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}