{
  "schemaVersion": 1,
  "videoId": "nZUPjFmtXxo",
  "videoUrl": "https://www.youtube.com/watch?v=nZUPjFmtXxo",
  "title": "Symbolic Execution of Math-Heavy Code in the Age of AI",
  "speakerCredit": "Neville Grech",
  "programmeTalkId": "dss26-d6c5dc970598c049",
  "language": "en",
  "captionType": "auto-generated",
  "humanAccuracyReviewed": false,
  "spokenAudioCompletenessVerified": false,
  "cueEndTimesAvailable": false,
  "source": {
    "provider": "YouTube",
    "method": "native browser transcript export",
    "sha256": "6bd7c14b04a83b68b010cb5397c5d8860cf6297cc696109b40c1304915e63cd8",
    "exportedAt": "2026-11-16T07:33:39.234094+00:00"
  },
  "qualityCaveats": [
    "Names, organizations, technical terms, code and numbers may be misrecognized.",
    "Source wording and disfluencies are preserved; no editorial corrections were applied.",
    "Export contains whole-second cue starts only; cue ends and subsecond timing are unavailable.",
    "A full caption-track export does not establish that every spoken word is represented.",
    "Automatic YouTube captions; not human-verified against the audio."
  ],
  "timingValidation": {
    "first_cue_start_seconds": 1,
    "last_cue_start_seconds": 1193,
    "video_duration_seconds": 1201,
    "seconds_after_last_cue_start": 8,
    "cue_start_gaps_over_30_seconds": [],
    "boundary_check": "pass",
    "criterion": "first cue <=15 seconds; last cue start within 30 seconds of player duration, allowing 1-second precision",
    "meaning": "Timing plausibility only; silence/applause or source caption omissions require audio review."
  },
  "segments": [
    {
      "index": 0,
      "start_seconds": 1,
      "source_timestamp": "0:01",
      "end_seconds": null,
      "text": "All right, thank you for the intro. I'm"
    },
    {
      "index": 1,
      "start_seconds": 4,
      "source_timestamp": "0:04",
      "end_seconds": null,
      "text": "Neville from DDUB. Just a little bit"
    },
    {
      "index": 2,
      "start_seconds": 6,
      "source_timestamp": "0:06",
      "end_seconds": null,
      "text": "about our organization. So, we're a"
    },
    {
      "index": 3,
      "start_seconds": 8,
      "source_timestamp": "0:08",
      "end_seconds": null,
      "text": "security company. We do audits. We're"
    },
    {
      "index": 4,
      "start_seconds": 11,
      "source_timestamp": "0:11",
      "end_seconds": null,
      "text": "very well known in the static analysis"
    },
    {
      "index": 5,
      "start_seconds": 13,
      "source_timestamp": "0:13",
      "end_seconds": null,
      "text": "and decompilation space. Uh we've built"
    },
    {
      "index": 6,
      "start_seconds": 15,
      "source_timestamp": "0:15",
      "end_seconds": null,
      "text": "a lot of products that use some of these"
    },
    {
      "index": 7,
      "start_seconds": 17,
      "source_timestamp": "0:17",
      "end_seconds": null,
      "text": "techniques. And as you'll see today,"
    },
    {
      "index": 8,
      "start_seconds": 19,
      "source_timestamp": "0:19",
      "end_seconds": null,
      "text": "we're also into symbolic execution,"
    },
    {
      "index": 9,
      "start_seconds": 21,
      "source_timestamp": "0:21",
      "end_seconds": null,
      "text": "which is a technique that underpins a"
    },
    {
      "index": 10,
      "start_seconds": 24,
      "source_timestamp": "0:24",
      "end_seconds": null,
      "text": "lot of the formal verification tools"
    },
    {
      "index": 11,
      "start_seconds": 26,
      "source_timestamp": "0:26",
      "end_seconds": null,
      "text": "that uh we use as a community."
    },
    {
      "index": 12,
      "start_seconds": 30,
      "source_timestamp": "0:30",
      "end_seconds": null,
      "text": "All right. So,"
    },
    {
      "index": 13,
      "start_seconds": 33,
      "source_timestamp": "0:33",
      "end_seconds": null,
      "text": "okay. So, what is symbolic execution?"
    },
    {
      "index": 14,
      "start_seconds": 38,
      "source_timestamp": "0:38",
      "end_seconds": null,
      "text": "All right. So, I'm going to use this"
    },
    {
      "index": 15,
      "start_seconds": 40,
      "source_timestamp": "0:40",
      "end_seconds": null,
      "text": "example as a running example. So, it's a"
    },
    {
      "index": 16,
      "start_seconds": 42,
      "source_timestamp": "0:42",
      "end_seconds": null,
      "text": "basically a transfer function. I made"
    },
    {
      "index": 17,
      "start_seconds": 45,
      "source_timestamp": "0:45",
      "end_seconds": null,
      "text": "this up. So, you will not see code as"
    },
    {
      "index": 18,
      "start_seconds": 47,
      "source_timestamp": "0:47",
      "end_seconds": null,
      "text": "bad in the wild as this. But effectively"
    },
    {
      "index": 19,
      "start_seconds": 50,
      "source_timestamp": "0:50",
      "end_seconds": null,
      "text": "in symbolic execution, you don't execute"
    },
    {
      "index": 20,
      "start_seconds": 52,
      "source_timestamp": "0:52",
      "end_seconds": null,
      "text": "the code using real inputs. Instead, you"
    },
    {
      "index": 21,
      "start_seconds": 56,
      "source_timestamp": "0:56",
      "end_seconds": null,
      "text": "execute the code using symbols. And"
    },
    {
      "index": 22,
      "start_seconds": 58,
      "source_timestamp": "0:58",
      "end_seconds": null,
      "text": "these symbols can represent all the"
    },
    {
      "index": 23,
      "start_seconds": 60,
      "source_timestamp": "1:00",
      "end_seconds": null,
      "text": "possible values uh of every variable"
    },
    {
      "index": 24,
      "start_seconds": 63,
      "source_timestamp": "1:03",
      "end_seconds": null,
      "text": "input that you have. Okay. So because of"
    },
    {
      "index": 25,
      "start_seconds": 67,
      "source_timestamp": "1:07",
      "end_seconds": null,
      "text": "this you end up um when you reach a"
    },
    {
      "index": 26,
      "start_seconds": 70,
      "source_timestamp": "1:10",
      "end_seconds": null,
      "text": "branch you end up forking meaning that"
    },
    {
      "index": 27,
      "start_seconds": 74,
      "source_timestamp": "1:14",
      "end_seconds": null,
      "text": "um you have a branch like an if"
    },
    {
      "index": 28,
      "start_seconds": 78,
      "source_timestamp": "1:18",
      "end_seconds": null,
      "text": "statement you got to execute the program"
    },
    {
      "index": 29,
      "start_seconds": 80,
      "source_timestamp": "1:20",
      "end_seconds": null,
      "text": "under both one side and the other side"
    },
    {
      "index": 30,
      "start_seconds": 82,
      "source_timestamp": "1:22",
      "end_seconds": null,
      "text": "concurrently."
    },
    {
      "index": 31,
      "start_seconds": 84,
      "source_timestamp": "1:24",
      "end_seconds": null,
      "text": "All right. So, uh, what this means is"
    },
    {
      "index": 32,
      "start_seconds": 86,
      "source_timestamp": "1:26",
      "end_seconds": null,
      "text": "that he will go down a lot of different"
    },
    {
      "index": 33,
      "start_seconds": 88,
      "source_timestamp": "1:28",
      "end_seconds": null,
      "text": "paths. Um, and I'm going to do a dry run"
    },
    {
      "index": 34,
      "start_seconds": 92,
      "source_timestamp": "1:32",
      "end_seconds": null,
      "text": "of a few of these paths."
    },
    {
      "index": 35,
      "start_seconds": 100,
      "source_timestamp": "1:40",
      "end_seconds": null,
      "text": "Okay. So, we start the program. All"
    },
    {
      "index": 36,
      "start_seconds": 103,
      "source_timestamp": "1:43",
      "end_seconds": null,
      "text": "right. So, we reach a condition. Reach a"
    },
    {
      "index": 37,
      "start_seconds": 106,
      "source_timestamp": "1:46",
      "end_seconds": null,
      "text": "condition. An if statement. This if"
    },
    {
      "index": 38,
      "start_seconds": 109,
      "source_timestamp": "1:49",
      "end_seconds": null,
      "text": "statement here checks, you know, that"
    },
    {
      "index": 39,
      "start_seconds": 110,
      "source_timestamp": "1:50",
      "end_seconds": null,
      "text": "the amount in this transfer is not over"
    },
    {
      "index": 40,
      "start_seconds": 113,
      "source_timestamp": "1:53",
      "end_seconds": null,
      "text": "a certain amount. Okay, at this point we"
    },
    {
      "index": 41,
      "start_seconds": 116,
      "source_timestamp": "1:56",
      "end_seconds": null,
      "text": "fork. So let's fork and go down one uh"
    },
    {
      "index": 42,
      "start_seconds": 119,
      "source_timestamp": "1:59",
      "end_seconds": null,
      "text": "branch. Okay, under this branch we're"
    },
    {
      "index": 43,
      "start_seconds": 122,
      "source_timestamp": "2:02",
      "end_seconds": null,
      "text": "going to assume that the condition is"
    },
    {
      "index": 44,
      "start_seconds": 125,
      "source_timestamp": "2:05",
      "end_seconds": null,
      "text": "false, meaning that the amount is less"
    },
    {
      "index": 45,
      "start_seconds": 128,
      "source_timestamp": "2:08",
      "end_seconds": null,
      "text": "than a th00and. Okay,"
    },
    {
      "index": 46,
      "start_seconds": 135,
      "source_timestamp": "2:15",
      "end_seconds": null,
      "text": "if we assume that we add that amount to"
    },
    {
      "index": 47,
      "start_seconds": 138,
      "source_timestamp": "2:18",
      "end_seconds": null,
      "text": "the so-called path condition. The path"
    },
    {
      "index": 48,
      "start_seconds": 140,
      "source_timestamp": "2:20",
      "end_seconds": null,
      "text": "condition is a set of conditions under"
    },
    {
      "index": 49,
      "start_seconds": 143,
      "source_timestamp": "2:23",
      "end_seconds": null,
      "text": "which the current under the current"
    },
    {
      "index": 50,
      "start_seconds": 145,
      "source_timestamp": "2:25",
      "end_seconds": null,
      "text": "execution that you must assume to be"
    },
    {
      "index": 51,
      "start_seconds": 147,
      "source_timestamp": "2:27",
      "end_seconds": null,
      "text": "true. Then you reach another condition."
    },
    {
      "index": 52,
      "start_seconds": 158,
      "source_timestamp": "2:38",
      "end_seconds": null,
      "text": "Okay, we reach another condition. This"
    },
    {
      "index": 53,
      "start_seconds": 160,
      "source_timestamp": "2:40",
      "end_seconds": null,
      "text": "is a condition about the uh the current"
    },
    {
      "index": 54,
      "start_seconds": 163,
      "source_timestamp": "2:43",
      "end_seconds": null,
      "text": "fee. In this case, we add also another"
    },
    {
      "index": 55,
      "start_seconds": 166,
      "source_timestamp": "2:46",
      "end_seconds": null,
      "text": "path condition. Then at the end, we need"
    },
    {
      "index": 56,
      "start_seconds": 169,
      "source_timestamp": "2:49",
      "end_seconds": null,
      "text": "to send these path conditions over to an"
    },
    {
      "index": 57,
      "start_seconds": 170,
      "source_timestamp": "2:50",
      "end_seconds": null,
      "text": "SMT solver. And the SMT solver will"
    },
    {
      "index": 58,
      "start_seconds": 173,
      "source_timestamp": "2:53",
      "end_seconds": null,
      "text": "check whether these conditions are"
    },
    {
      "index": 59,
      "start_seconds": 176,
      "source_timestamp": "2:56",
      "end_seconds": null,
      "text": "consistent with each other and it will"
    },
    {
      "index": 60,
      "start_seconds": 177,
      "source_timestamp": "2:57",
      "end_seconds": null,
      "text": "give us either a satisfiable assignment"
    },
    {
      "index": 61,
      "start_seconds": 179,
      "source_timestamp": "2:59",
      "end_seconds": null,
      "text": "SAT or unsatisfiable. Okay, a lot of"
    },
    {
      "index": 62,
      "start_seconds": 183,
      "source_timestamp": "3:03",
      "end_seconds": null,
      "text": "people use Z3 which is an external"
    },
    {
      "index": 63,
      "start_seconds": 185,
      "source_timestamp": "3:05",
      "end_seconds": null,
      "text": "solver but you can also write your own"
    },
    {
      "index": 64,
      "start_seconds": 188,
      "source_timestamp": "3:08",
      "end_seconds": null,
      "text": "solver. It's just difficult."
    },
    {
      "index": 65,
      "start_seconds": 191,
      "source_timestamp": "3:11",
      "end_seconds": null,
      "text": "So in this case turns out that this is"
    },
    {
      "index": 66,
      "start_seconds": 193,
      "source_timestamp": "3:13",
      "end_seconds": null,
      "text": "ansat. Okay. So we terminate the"
    },
    {
      "index": 67,
      "start_seconds": 196,
      "source_timestamp": "3:16",
      "end_seconds": null,
      "text": "execution here. We need to repeat the"
    },
    {
      "index": 68,
      "start_seconds": 199,
      "source_timestamp": "3:19",
      "end_seconds": null,
      "text": "same process to all possible paths. In"
    },
    {
      "index": 69,
      "start_seconds": 201,
      "source_timestamp": "3:21",
      "end_seconds": null,
      "text": "this case the program only contains"
    },
    {
      "index": 70,
      "start_seconds": 203,
      "source_timestamp": "3:23",
      "end_seconds": null,
      "text": "three valid paths. But as you can"
    },
    {
      "index": 71,
      "start_seconds": 206,
      "source_timestamp": "3:26",
      "end_seconds": null,
      "text": "imagine, real programs are going to have"
    },
    {
      "index": 72,
      "start_seconds": 208,
      "source_timestamp": "3:28",
      "end_seconds": null,
      "text": "a lot more paths because if you have"
    },
    {
      "index": 73,
      "start_seconds": 211,
      "source_timestamp": "3:31",
      "end_seconds": null,
      "text": "let's say 45 branches"
    },
    {
      "index": 74,
      "start_seconds": 214,
      "source_timestamp": "3:34",
      "end_seconds": null,
      "text": "and each branch has a branching factor"
    },
    {
      "index": 75,
      "start_seconds": 216,
      "source_timestamp": "3:36",
      "end_seconds": null,
      "text": "of like 1.5 meaning that you know one"
    },
    {
      "index": 76,
      "start_seconds": 219,
      "source_timestamp": "3:39",
      "end_seconds": null,
      "text": "side succeeds the other side has a 50%"
    },
    {
      "index": 77,
      "start_seconds": 222,
      "source_timestamp": "3:42",
      "end_seconds": null,
      "text": "of continuing you will have 84 million"
    },
    {
      "index": 78,
      "start_seconds": 226,
      "source_timestamp": "3:46",
      "end_seconds": null,
      "text": "paths. Okay, which is which is a lot of"
    },
    {
      "index": 79,
      "start_seconds": 229,
      "source_timestamp": "3:49",
      "end_seconds": null,
      "text": "paths and you know a single query with"
    },
    {
      "index": 80,
      "start_seconds": 232,
      "source_timestamp": "3:52",
      "end_seconds": null,
      "text": "an SMT solver it can take minutes"
    },
    {
      "index": 81,
      "start_seconds": 235,
      "source_timestamp": "3:55",
      "end_seconds": null,
      "text": "potentially"
    },
    {
      "index": 82,
      "start_seconds": 237,
      "source_timestamp": "3:57",
      "end_seconds": null,
      "text": "and actually analyzing D5 protocols is"
    },
    {
      "index": 83,
      "start_seconds": 240,
      "source_timestamp": "4:00",
      "end_seconds": null,
      "text": "harder. Why is it harder?"
    },
    {
      "index": 84,
      "start_seconds": 248,
      "source_timestamp": "4:08",
      "end_seconds": null,
      "text": "It's harder because you find a lot of"
    },
    {
      "index": 85,
      "start_seconds": 250,
      "source_timestamp": "4:10",
      "end_seconds": null,
      "text": "code in math libraries which uh which"
    },
    {
      "index": 86,
      "start_seconds": 254,
      "source_timestamp": "4:14",
      "end_seconds": null,
      "text": "combines different kinds of algebbras"
    },
    {
      "index": 87,
      "start_seconds": 257,
      "source_timestamp": "4:17",
      "end_seconds": null,
      "text": "meaning that it uses both for instance"
    },
    {
      "index": 88,
      "start_seconds": 259,
      "source_timestamp": "4:19",
      "end_seconds": null,
      "text": "addition, multiplication and subtraction"
    },
    {
      "index": 89,
      "start_seconds": 262,
      "source_timestamp": "4:22",
      "end_seconds": null,
      "text": "which you need certain techniques to"
    },
    {
      "index": 90,
      "start_seconds": 264,
      "source_timestamp": "4:24",
      "end_seconds": null,
      "text": "reason about and also it uses you know"
    },
    {
      "index": 91,
      "start_seconds": 267,
      "source_timestamp": "4:27",
      "end_seconds": null,
      "text": "shifts uh and or or bitwise operations"
    },
    {
      "index": 92,
      "start_seconds": 271,
      "source_timestamp": "4:31",
      "end_seconds": null,
      "text": "which require different techniques. And"
    },
    {
      "index": 93,
      "start_seconds": 274,
      "source_timestamp": "4:34",
      "end_seconds": null,
      "text": "so if you go down and use one kind of"
    },
    {
      "index": 94,
      "start_seconds": 277,
      "source_timestamp": "4:37",
      "end_seconds": null,
      "text": "techniques, you need to convert one kind"
    },
    {
      "index": 95,
      "start_seconds": 280,
      "source_timestamp": "4:40",
      "end_seconds": null,
      "text": "of math into another kind and that's"
    },
    {
      "index": 96,
      "start_seconds": 283,
      "source_timestamp": "4:43",
      "end_seconds": null,
      "text": "creates a lot of complexity. This is an"
    },
    {
      "index": 97,
      "start_seconds": 286,
      "source_timestamp": "4:46",
      "end_seconds": null,
      "text": "example from unis swap v3. In this case"
    },
    {
      "index": 98,
      "start_seconds": 289,
      "source_timestamp": "4:49",
      "end_seconds": null,
      "text": "it does 512 bit multiplication"
    },
    {
      "index": 99,
      "start_seconds": 293,
      "source_timestamp": "4:53",
      "end_seconds": null,
      "text": "followed by division."
    },
    {
      "index": 100,
      "start_seconds": 295,
      "source_timestamp": "4:55",
      "end_seconds": null,
      "text": "All right. And then on top of that you"
    },
    {
      "index": 101,
      "start_seconds": 299,
      "source_timestamp": "4:59",
      "end_seconds": null,
      "text": "also get complexity."
    },
    {
      "index": 102,
      "start_seconds": 301,
      "source_timestamp": "5:01",
      "end_seconds": null,
      "text": "Let me show you an example."
    },
    {
      "index": 103,
      "start_seconds": 308,
      "source_timestamp": "5:08",
      "end_seconds": null,
      "text": "So you get additional complexity. This"
    },
    {
      "index": 104,
      "start_seconds": 310,
      "source_timestamp": "5:10",
      "end_seconds": null,
      "text": "is another example. This is also from"
    },
    {
      "index": 105,
      "start_seconds": 312,
      "source_timestamp": "5:12",
      "end_seconds": null,
      "text": "unis swap. But in this case, it's a"
    },
    {
      "index": 106,
      "start_seconds": 314,
      "source_timestamp": "5:14",
      "end_seconds": null,
      "text": "function that helps you compute the"
    },
    {
      "index": 107,
      "start_seconds": 316,
      "source_timestamp": "5:16",
      "end_seconds": null,
      "text": "price of an asset based on the liquidity"
    },
    {
      "index": 108,
      "start_seconds": 318,
      "source_timestamp": "5:18",
      "end_seconds": null,
      "text": "in a in a particular tick. In this case,"
    },
    {
      "index": 109,
      "start_seconds": 321,
      "source_timestamp": "5:21",
      "end_seconds": null,
      "text": "you get like 20 possible if statements."
    },
    {
      "index": 110,
      "start_seconds": 324,
      "source_timestamp": "5:24",
      "end_seconds": null,
      "text": "And you know, all branches are mostly"
    },
    {
      "index": 111,
      "start_seconds": 326,
      "source_timestamp": "5:26",
      "end_seconds": null,
      "text": "feasible. And so you get once you enter"
    },
    {
      "index": 112,
      "start_seconds": 329,
      "source_timestamp": "5:29",
      "end_seconds": null,
      "text": "this function, you multiply the number"
    },
    {
      "index": 113,
      "start_seconds": 331,
      "source_timestamp": "5:31",
      "end_seconds": null,
      "text": "of pads by around 1 million 2 to the 20."
    },
    {
      "index": 114,
      "start_seconds": 335,
      "source_timestamp": "5:35",
      "end_seconds": null,
      "text": "Okay. And the correctness of these D5"
    },
    {
      "index": 115,
      "start_seconds": 337,
      "source_timestamp": "5:37",
      "end_seconds": null,
      "text": "protocols really relies on code like"
    },
    {
      "index": 116,
      "start_seconds": 340,
      "source_timestamp": "5:40",
      "end_seconds": null,
      "text": "this. A lot of tools will not support"
    },
    {
      "index": 117,
      "start_seconds": 343,
      "source_timestamp": "5:43",
      "end_seconds": null,
      "text": "code like this. A lot of provers will"
    },
    {
      "index": 118,
      "start_seconds": 345,
      "source_timestamp": "5:45",
      "end_seconds": null,
      "text": "will choke. So what do we do?"
    },
    {
      "index": 119,
      "start_seconds": 348,
      "source_timestamp": "5:48",
      "end_seconds": null,
      "text": "So we've devised a series of techniques."
    },
    {
      "index": 120,
      "start_seconds": 352,
      "source_timestamp": "5:52",
      "end_seconds": null,
      "text": "Okay, some of these techniques are"
    },
    {
      "index": 121,
      "start_seconds": 353,
      "source_timestamp": "5:53",
      "end_seconds": null,
      "text": "original, some of them are appropriated."
    },
    {
      "index": 122,
      "start_seconds": 356,
      "source_timestamp": "5:56",
      "end_seconds": null,
      "text": "Okay. So the first thing we do, okay, uh"
    },
    {
      "index": 123,
      "start_seconds": 360,
      "source_timestamp": "6:00",
      "end_seconds": null,
      "text": "we are known in the decompiler space."
    },
    {
      "index": 124,
      "start_seconds": 362,
      "source_timestamp": "6:02",
      "end_seconds": null,
      "text": "We've written a lot of techniques to try"
    },
    {
      "index": 125,
      "start_seconds": 364,
      "source_timestamp": "6:04",
      "end_seconds": null,
      "text": "to lift code at a level of"
    },
    {
      "index": 126,
      "start_seconds": 366,
      "source_timestamp": "6:06",
      "end_seconds": null,
      "text": "representation higher than the bite"
    },
    {
      "index": 127,
      "start_seconds": 369,
      "source_timestamp": "6:09",
      "end_seconds": null,
      "text": "code. Okay, that's in the DOB"
    },
    {
      "index": 128,
      "start_seconds": 371,
      "source_timestamp": "6:11",
      "end_seconds": null,
      "text": "decompiler. We've used some of those"
    },
    {
      "index": 129,
      "start_seconds": 373,
      "source_timestamp": "6:13",
      "end_seconds": null,
      "text": "same techniques to lift low-level code"
    },
    {
      "index": 130,
      "start_seconds": 377,
      "source_timestamp": "6:17",
      "end_seconds": null,
      "text": "also to a higher level presentation. The"
    },
    {
      "index": 131,
      "start_seconds": 380,
      "source_timestamp": "6:20",
      "end_seconds": null,
      "text": "screenshot uh shows a simple program"
    },
    {
      "index": 132,
      "start_seconds": 383,
      "source_timestamp": "6:23",
      "end_seconds": null,
      "text": "which has been compiled and it doesn't"
    },
    {
      "index": 133,
      "start_seconds": 385,
      "source_timestamp": "6:25",
      "end_seconds": null,
      "text": "do anything special. It just copies a"
    },
    {
      "index": 134,
      "start_seconds": 387,
      "source_timestamp": "6:27",
      "end_seconds": null,
      "text": "strct around call data storage and that"
    },
    {
      "index": 135,
      "start_seconds": 391,
      "source_timestamp": "6:31",
      "end_seconds": null,
      "text": "sort of thing and does some arithmetic."
    },
    {
      "index": 136,
      "start_seconds": 394,
      "source_timestamp": "6:34",
      "end_seconds": null,
      "text": "So for instance the uh code in green is"
    },
    {
      "index": 137,
      "start_seconds": 398,
      "source_timestamp": "6:38",
      "end_seconds": null,
      "text": "you know like check arithmetic."
    },
    {
      "index": 138,
      "start_seconds": 400,
      "source_timestamp": "6:40",
      "end_seconds": null,
      "text": "It really translates to like three or"
    },
    {
      "index": 139,
      "start_seconds": 403,
      "source_timestamp": "6:43",
      "end_seconds": null,
      "text": "four mathematical operation. So this is"
    },
    {
      "index": 140,
      "start_seconds": 406,
      "source_timestamp": "6:46",
      "end_seconds": null,
      "text": "the sort of thing we do. There's a lot"
    },
    {
      "index": 141,
      "start_seconds": 408,
      "source_timestamp": "6:48",
      "end_seconds": null,
      "text": "of techniques which go into uh doing"
    },
    {
      "index": 142,
      "start_seconds": 410,
      "source_timestamp": "6:50",
      "end_seconds": null,
      "text": "this convert lifting to a higher level"
    },
    {
      "index": 143,
      "start_seconds": 413,
      "source_timestamp": "6:53",
      "end_seconds": null,
      "text": "representation. Second thing we do is we"
    },
    {
      "index": 144,
      "start_seconds": 417,
      "source_timestamp": "6:57",
      "end_seconds": null,
      "text": "undo the optimizations."
    },
    {
      "index": 145,
      "start_seconds": 419,
      "source_timestamp": "6:59",
      "end_seconds": null,
      "text": "Now this is easier said than done and"
    },
    {
      "index": 146,
      "start_seconds": 422,
      "source_timestamp": "7:02",
      "end_seconds": null,
      "text": "also you might question why would you"
    },
    {
      "index": 147,
      "start_seconds": 423,
      "source_timestamp": "7:03",
      "end_seconds": null,
      "text": "want to undo the optimizations? I mean"
    },
    {
      "index": 148,
      "start_seconds": 426,
      "source_timestamp": "7:06",
      "end_seconds": null,
      "text": "they're there for a reason. Well, the"
    },
    {
      "index": 149,
      "start_seconds": 428,
      "source_timestamp": "7:08",
      "end_seconds": null,
      "text": "problem with optimizations is that the"
    },
    {
      "index": 150,
      "start_seconds": 431,
      "source_timestamp": "7:11",
      "end_seconds": null,
      "text": "uh library auditor authors, they have"
    },
    {
      "index": 151,
      "start_seconds": 434,
      "source_timestamp": "7:14",
      "end_seconds": null,
      "text": "exploited the fact that they're using a"
    },
    {
      "index": 152,
      "start_seconds": 436,
      "source_timestamp": "7:16",
      "end_seconds": null,
      "text": "a machine, you know, with with bits and"
    },
    {
      "index": 153,
      "start_seconds": 439,
      "source_timestamp": "7:19",
      "end_seconds": null,
      "text": "that sort of thing. And by exploiting"
    },
    {
      "index": 154,
      "start_seconds": 441,
      "source_timestamp": "7:21",
      "end_seconds": null,
      "text": "this, they can get better performance"
    },
    {
      "index": 155,
      "start_seconds": 443,
      "source_timestamp": "7:23",
      "end_seconds": null,
      "text": "and in some cases they can get also"
    },
    {
      "index": 156,
      "start_seconds": 445,
      "source_timestamp": "7:25",
      "end_seconds": null,
      "text": "better um better accuracy, you know, can"
    },
    {
      "index": 157,
      "start_seconds": 449,
      "source_timestamp": "7:29",
      "end_seconds": null,
      "text": "they can utilize uh the bits better."
    },
    {
      "index": 158,
      "start_seconds": 454,
      "source_timestamp": "7:34",
      "end_seconds": null,
      "text": "Um but unfortunately these also"
    },
    {
      "index": 159,
      "start_seconds": 456,
      "source_timestamp": "7:36",
      "end_seconds": null,
      "text": "introduced these mixed algebra um"
    },
    {
      "index": 160,
      "start_seconds": 459,
      "source_timestamp": "7:39",
      "end_seconds": null,
      "text": "expressions."
    },
    {
      "index": 161,
      "start_seconds": 461,
      "source_timestamp": "7:41",
      "end_seconds": null,
      "text": "So we do one of two things when we see"
    },
    {
      "index": 162,
      "start_seconds": 464,
      "source_timestamp": "7:44",
      "end_seconds": null,
      "text": "an expression which is problematic. We"
    },
    {
      "index": 163,
      "start_seconds": 466,
      "source_timestamp": "7:46",
      "end_seconds": null,
      "text": "have devised algorithms to either"
    },
    {
      "index": 164,
      "start_seconds": 468,
      "source_timestamp": "7:48",
      "end_seconds": null,
      "text": "convert these to idealized mathematical"
    },
    {
      "index": 165,
      "start_seconds": 471,
      "source_timestamp": "7:51",
      "end_seconds": null,
      "text": "representations such as in the first"
    },
    {
      "index": 166,
      "start_seconds": 473,
      "source_timestamp": "7:53",
      "end_seconds": null,
      "text": "example that's the min function. Second"
    },
    {
      "index": 167,
      "start_seconds": 476,
      "source_timestamp": "7:56",
      "end_seconds": null,
      "text": "function is the average function. It's"
    },
    {
      "index": 168,
      "start_seconds": 479,
      "source_timestamp": "7:59",
      "end_seconds": null,
      "text": "simpler to reason about than the one on"
    },
    {
      "index": 169,
      "start_seconds": 482,
      "source_timestamp": "8:02",
      "end_seconds": null,
      "text": "the left hand side. Or another technique"
    },
    {
      "index": 170,
      "start_seconds": 485,
      "source_timestamp": "8:05",
      "end_seconds": null,
      "text": "is to convert for instance a bits"
    },
    {
      "index": 171,
      "start_seconds": 488,
      "source_timestamp": "8:08",
      "end_seconds": null,
      "text": "operation into an operation that"
    },
    {
      "index": 172,
      "start_seconds": 491,
      "source_timestamp": "8:11",
      "end_seconds": null,
      "text": "operates on data structures. So the"
    },
    {
      "index": 173,
      "start_seconds": 494,
      "source_timestamp": "8:14",
      "end_seconds": null,
      "text": "operation on the right hand side is a"
    },
    {
      "index": 174,
      "start_seconds": 496,
      "source_timestamp": "8:16",
      "end_seconds": null,
      "text": "slice of bits and you want to be careful"
    },
    {
      "index": 175,
      "start_seconds": 499,
      "source_timestamp": "8:19",
      "end_seconds": null,
      "text": "you know the algorithms need to be"
    },
    {
      "index": 176,
      "start_seconds": 500,
      "source_timestamp": "8:20",
      "end_seconds": null,
      "text": "careful to distinguish between the two."
    },
    {
      "index": 177,
      "start_seconds": 503,
      "source_timestamp": "8:23",
      "end_seconds": null,
      "text": "Okay, by the way, a lot of optimizations"
    },
    {
      "index": 178,
      "start_seconds": 506,
      "source_timestamp": "8:26",
      "end_seconds": null,
      "text": "you can read about them in this book,"
    },
    {
      "index": 179,
      "start_seconds": 508,
      "source_timestamp": "8:28",
      "end_seconds": null,
      "text": "but this was written for C. It's a good"
    },
    {
      "index": 180,
      "start_seconds": 510,
      "source_timestamp": "8:30",
      "end_seconds": null,
      "text": "book."
    },
    {
      "index": 181,
      "start_seconds": 515,
      "source_timestamp": "8:35",
      "end_seconds": null,
      "text": "All right. So, another technique, it's"
    },
    {
      "index": 182,
      "start_seconds": 517,
      "source_timestamp": "8:37",
      "end_seconds": null,
      "text": "to write your own SMT solver. Now, this"
    },
    {
      "index": 183,
      "start_seconds": 521,
      "source_timestamp": "8:41",
      "end_seconds": null,
      "text": "is uh not as simple as I'm saying. Um,"
    },
    {
      "index": 184,
      "start_seconds": 525,
      "source_timestamp": "8:45",
      "end_seconds": null,
      "text": "so you don't need to write your own SMT"
    },
    {
      "index": 185,
      "start_seconds": 527,
      "source_timestamp": "8:47",
      "end_seconds": null,
      "text": "solver. Zedri does a very good job, but"
    },
    {
      "index": 186,
      "start_seconds": 529,
      "source_timestamp": "8:49",
      "end_seconds": null,
      "text": "it is designed to give you either a"
    },
    {
      "index": 187,
      "start_seconds": 531,
      "source_timestamp": "8:51",
      "end_seconds": null,
      "text": "satisfiable assignment or unsatisfiable."
    },
    {
      "index": 188,
      "start_seconds": 534,
      "source_timestamp": "8:54",
      "end_seconds": null,
      "text": "And to get a satisfiable assignment that"
    },
    {
      "index": 189,
      "start_seconds": 538,
      "source_timestamp": "8:58",
      "end_seconds": null,
      "text": "um it requires more computation than you"
    },
    {
      "index": 190,
      "start_seconds": 541,
      "source_timestamp": "9:01",
      "end_seconds": null,
      "text": "might need. Instead, during symbolic"
    },
    {
      "index": 191,
      "start_seconds": 544,
      "source_timestamp": "9:04",
      "end_seconds": null,
      "text": "execution, often times you're more"
    },
    {
      "index": 192,
      "start_seconds": 546,
      "source_timestamp": "9:06",
      "end_seconds": null,
      "text": "interested to know whether a uh path"
    },
    {
      "index": 193,
      "start_seconds": 549,
      "source_timestamp": "9:09",
      "end_seconds": null,
      "text": "condition is completely unsatisfiable or"
    },
    {
      "index": 194,
      "start_seconds": 553,
      "source_timestamp": "9:13",
      "end_seconds": null,
      "text": "whether it could possibly be unsat"
    },
    {
      "index": 195,
      "start_seconds": 555,
      "source_timestamp": "9:15",
      "end_seconds": null,
      "text": "possibly be satisfiable. And so writing"
    },
    {
      "index": 196,
      "start_seconds": 558,
      "source_timestamp": "9:18",
      "end_seconds": null,
      "text": "a solver like that is simpler and also"
    },
    {
      "index": 197,
      "start_seconds": 562,
      "source_timestamp": "9:22",
      "end_seconds": null,
      "text": "the algorithms that you can pick and"
    },
    {
      "index": 198,
      "start_seconds": 564,
      "source_timestamp": "9:24",
      "end_seconds": null,
      "text": "choose into the solver. You can make the"
    },
    {
      "index": 199,
      "start_seconds": 567,
      "source_timestamp": "9:27",
      "end_seconds": null,
      "text": "solver be linear time and you can"
    },
    {
      "index": 200,
      "start_seconds": 569,
      "source_timestamp": "9:29",
      "end_seconds": null,
      "text": "utilize techniques from computer aided"
    },
    {
      "index": 201,
      "start_seconds": 573,
      "source_timestamp": "9:33",
      "end_seconds": null,
      "text": "algebra systems which are faster and I"
    },
    {
      "index": 202,
      "start_seconds": 576,
      "source_timestamp": "9:36",
      "end_seconds": null,
      "text": "don't know some of the these techniques"
    },
    {
      "index": 203,
      "start_seconds": 577,
      "source_timestamp": "9:37",
      "end_seconds": null,
      "text": "are also used in Z3 but you know linear"
    },
    {
      "index": 204,
      "start_seconds": 580,
      "source_timestamp": "9:40",
      "end_seconds": null,
      "text": "time versus double exponential but"
    },
    {
      "index": 205,
      "start_seconds": 584,
      "source_timestamp": "9:44",
      "end_seconds": null,
      "text": "there's a caveat at the end of the day"
    },
    {
      "index": 206,
      "start_seconds": 586,
      "source_timestamp": "9:46",
      "end_seconds": null,
      "text": "you still need to use Z3 in some cases."
    },
    {
      "index": 207,
      "start_seconds": 590,
      "source_timestamp": "9:50",
      "end_seconds": null,
      "text": "All right."
    },
    {
      "index": 208,
      "start_seconds": 594,
      "source_timestamp": "9:54",
      "end_seconds": null,
      "text": "Another technique this is actually"
    },
    {
      "index": 209,
      "start_seconds": 597,
      "source_timestamp": "9:57",
      "end_seconds": null,
      "text": "appropriated. A lot of uh symbolic"
    },
    {
      "index": 210,
      "start_seconds": 599,
      "source_timestamp": "9:59",
      "end_seconds": null,
      "text": "execution engines they do summaries."
    },
    {
      "index": 211,
      "start_seconds": 602,
      "source_timestamp": "10:02",
      "end_seconds": null,
      "text": "Okay. We also do summaries in our"
    },
    {
      "index": 212,
      "start_seconds": 604,
      "source_timestamp": "10:04",
      "end_seconds": null,
      "text": "symbolic execution engine. So summaries"
    },
    {
      "index": 213,
      "start_seconds": 607,
      "source_timestamp": "10:07",
      "end_seconds": null,
      "text": "you are actually encoding instead of"
    },
    {
      "index": 214,
      "start_seconds": 609,
      "source_timestamp": "10:09",
      "end_seconds": null,
      "text": "like um forcing the engine to execute a"
    },
    {
      "index": 215,
      "start_seconds": 613,
      "source_timestamp": "10:13",
      "end_seconds": null,
      "text": "lot of difficult and complex code you"
    },
    {
      "index": 216,
      "start_seconds": 616,
      "source_timestamp": "10:16",
      "end_seconds": null,
      "text": "either reimplement it in a simpler way"
    },
    {
      "index": 217,
      "start_seconds": 618,
      "source_timestamp": "10:18",
      "end_seconds": null,
      "text": "using idealized mathematical operations"
    },
    {
      "index": 218,
      "start_seconds": 621,
      "source_timestamp": "10:21",
      "end_seconds": null,
      "text": "or you just put in there you know the"
    },
    {
      "index": 219,
      "start_seconds": 624,
      "source_timestamp": "10:24",
      "end_seconds": null,
      "text": "additional path conditions that you"
    },
    {
      "index": 220,
      "start_seconds": 626,
      "source_timestamp": "10:26",
      "end_seconds": null,
      "text": "would need to add. Okay. So the way we"
    },
    {
      "index": 221,
      "start_seconds": 630,
      "source_timestamp": "10:30",
      "end_seconds": null,
      "text": "uh wrote our summaries is we went"
    },
    {
      "index": 222,
      "start_seconds": 633,
      "source_timestamp": "10:33",
      "end_seconds": null,
      "text": "through a lot of defi and math libraries"
    },
    {
      "index": 223,
      "start_seconds": 635,
      "source_timestamp": "10:35",
      "end_seconds": null,
      "text": "and we asked AI to write summaries okay"
    },
    {
      "index": 224,
      "start_seconds": 639,
      "source_timestamp": "10:39",
      "end_seconds": null,
      "text": "and our symbolic execution engines have"
    },
    {
      "index": 225,
      "start_seconds": 641,
      "source_timestamp": "10:41",
      "end_seconds": null,
      "text": "has a lot of summaries in it now that"
    },
    {
      "index": 226,
      "start_seconds": 644,
      "source_timestamp": "10:44",
      "end_seconds": null,
      "text": "matches to specific libraries and"
    },
    {
      "index": 227,
      "start_seconds": 647,
      "source_timestamp": "10:47",
      "end_seconds": null,
      "text": "obviously we don't trust the AI so we uh"
    },
    {
      "index": 228,
      "start_seconds": 651,
      "source_timestamp": "10:51",
      "end_seconds": null,
      "text": "we use fuzzing to you know ver validate"
    },
    {
      "index": 229,
      "start_seconds": 655,
      "source_timestamp": "10:55",
      "end_seconds": null,
      "text": "these summaries against the actual code"
    },
    {
      "index": 230,
      "start_seconds": 657,
      "source_timestamp": "10:57",
      "end_seconds": null,
      "text": "now you might be thinking why use"
    },
    {
      "index": 231,
      "start_seconds": 659,
      "source_timestamp": "10:59",
      "end_seconds": null,
      "text": "fuzzing when you can use symbolic"
    },
    {
      "index": 232,
      "start_seconds": 661,
      "source_timestamp": "11:01",
      "end_seconds": null,
      "text": "execution. Well, for small functions,"
    },
    {
      "index": 233,
      "start_seconds": 664,
      "source_timestamp": "11:04",
      "end_seconds": null,
      "text": "small pure functions um that don't do"
    },
    {
      "index": 234,
      "start_seconds": 668,
      "source_timestamp": "11:08",
      "end_seconds": null,
      "text": "much. Okay, so these are mathematical"
    },
    {
      "index": 235,
      "start_seconds": 670,
      "source_timestamp": "11:10",
      "end_seconds": null,
      "text": "functions, fuzzers can actually do a"
    },
    {
      "index": 236,
      "start_seconds": 672,
      "source_timestamp": "11:12",
      "end_seconds": null,
      "text": "good job at disproving you and so it's"
    },
    {
      "index": 237,
      "start_seconds": 675,
      "source_timestamp": "11:15",
      "end_seconds": null,
      "text": "easier."
    },
    {
      "index": 238,
      "start_seconds": 678,
      "source_timestamp": "11:18",
      "end_seconds": null,
      "text": "And finally, if you accept the fact that"
    },
    {
      "index": 239,
      "start_seconds": 681,
      "source_timestamp": "11:21",
      "end_seconds": null,
      "text": "your symbolic execution engine is going"
    },
    {
      "index": 240,
      "start_seconds": 684,
      "source_timestamp": "11:24",
      "end_seconds": null,
      "text": "to be more scalable, but as a as a"
    },
    {
      "index": 241,
      "start_seconds": 686,
      "source_timestamp": "11:26",
      "end_seconds": null,
      "text": "result, you might get a few more false"
    },
    {
      "index": 242,
      "start_seconds": 689,
      "source_timestamp": "11:29",
      "end_seconds": null,
      "text": "positives."
    },
    {
      "index": 243,
      "start_seconds": 690,
      "source_timestamp": "11:30",
      "end_seconds": null,
      "text": "Well, you can have an a agent in the"
    },
    {
      "index": 244,
      "start_seconds": 693,
      "source_timestamp": "11:33",
      "end_seconds": null,
      "text": "loop that verifies, you know, whether"
    },
    {
      "index": 245,
      "start_seconds": 695,
      "source_timestamp": "11:35",
      "end_seconds": null,
      "text": "whether a finding is true positive or"
    },
    {
      "index": 246,
      "start_seconds": 697,
      "source_timestamp": "11:37",
      "end_seconds": null,
      "text": "false positive. You want to be using"
    },
    {
      "index": 247,
      "start_seconds": 700,
      "source_timestamp": "11:40",
      "end_seconds": null,
      "text": "frontier models, but frontier models are"
    },
    {
      "index": 248,
      "start_seconds": 702,
      "source_timestamp": "11:42",
      "end_seconds": null,
      "text": "very expensive. But how can you get the"
    },
    {
      "index": 249,
      "start_seconds": 704,
      "source_timestamp": "11:44",
      "end_seconds": null,
      "text": "cost down of to these frontier models? A"
    },
    {
      "index": 250,
      "start_seconds": 707,
      "source_timestamp": "11:47",
      "end_seconds": null,
      "text": "lot of tools when they give you a"
    },
    {
      "index": 251,
      "start_seconds": 709,
      "source_timestamp": "11:49",
      "end_seconds": null,
      "text": "finding they produce just gibberish or I"
    },
    {
      "index": 252,
      "start_seconds": 712,
      "source_timestamp": "11:52",
      "end_seconds": null,
      "text": "mean an expert can interpret it but it's"
    },
    {
      "index": 253,
      "start_seconds": 714,
      "source_timestamp": "11:54",
      "end_seconds": null,
      "text": "like really long okay and really long"
    },
    {
      "index": 254,
      "start_seconds": 717,
      "source_timestamp": "11:57",
      "end_seconds": null,
      "text": "traces through bite code and through"
    },
    {
      "index": 255,
      "start_seconds": 720,
      "source_timestamp": "12:00",
      "end_seconds": null,
      "text": "memory operations and that sort of thing"
    },
    {
      "index": 256,
      "start_seconds": 722,
      "source_timestamp": "12:02",
      "end_seconds": null,
      "text": "and it's a waste of tokens to go through"
    },
    {
      "index": 257,
      "start_seconds": 724,
      "source_timestamp": "12:04",
      "end_seconds": null,
      "text": "that. So a lot of you can you should"
    },
    {
      "index": 258,
      "start_seconds": 727,
      "source_timestamp": "12:07",
      "end_seconds": null,
      "text": "dedicate a lot of engineering to try to"
    },
    {
      "index": 259,
      "start_seconds": 730,
      "source_timestamp": "12:10",
      "end_seconds": null,
      "text": "decipher that and get a more highlevel"
    },
    {
      "index": 260,
      "start_seconds": 733,
      "source_timestamp": "12:13",
      "end_seconds": null,
      "text": "output from your tool so that the AI"
    },
    {
      "index": 261,
      "start_seconds": 735,
      "source_timestamp": "12:15",
      "end_seconds": null,
      "text": "agent does not need to spend a lot of"
    },
    {
      "index": 262,
      "start_seconds": 738,
      "source_timestamp": "12:18",
      "end_seconds": null,
      "text": "tokens on this and it's going to perform"
    },
    {
      "index": 263,
      "start_seconds": 740,
      "source_timestamp": "12:20",
      "end_seconds": null,
      "text": "better and then you can use a better"
    },
    {
      "index": 264,
      "start_seconds": 744,
      "source_timestamp": "12:24",
      "end_seconds": null,
      "text": "frontier model."
    },
    {
      "index": 265,
      "start_seconds": 746,
      "source_timestamp": "12:26",
      "end_seconds": null,
      "text": "This is actual input output from from"
    },
    {
      "index": 266,
      "start_seconds": 748,
      "source_timestamp": "12:28",
      "end_seconds": null,
      "text": "our tool. Okay. So everything maps back"
    },
    {
      "index": 267,
      "start_seconds": 751,
      "source_timestamp": "12:31",
      "end_seconds": null,
      "text": "to source code operations. Okay. It's"
    },
    {
      "index": 268,
      "start_seconds": 754,
      "source_timestamp": "12:34",
      "end_seconds": null,
      "text": "it's in a familiar format. That's also"
    },
    {
      "index": 269,
      "start_seconds": 756,
      "source_timestamp": "12:36",
      "end_seconds": null,
      "text": "another thing that you know they uh tell"
    },
    {
      "index": 270,
      "start_seconds": 759,
      "source_timestamp": "12:39",
      "end_seconds": null,
      "text": "you you know how to integrate with AI"
    },
    {
      "index": 271,
      "start_seconds": 761,
      "source_timestamp": "12:41",
      "end_seconds": null,
      "text": "models. Well, you make the output"
    },
    {
      "index": 272,
      "start_seconds": 764,
      "source_timestamp": "12:44",
      "end_seconds": null,
      "text": "familiar to them as well."
    },
    {
      "index": 273,
      "start_seconds": 770,
      "source_timestamp": "12:50",
      "end_seconds": null,
      "text": "Okay, let's look at some results."
    },
    {
      "index": 274,
      "start_seconds": 774,
      "source_timestamp": "12:54",
      "end_seconds": null,
      "text": "So, we've implemented this tool."
    },
    {
      "index": 275,
      "start_seconds": 779,
      "source_timestamp": "12:59",
      "end_seconds": null,
      "text": "It's symbolic execution engine and we"
    },
    {
      "index": 276,
      "start_seconds": 781,
      "source_timestamp": "13:01",
      "end_seconds": null,
      "text": "compared it against HEVM and Halmos."
    },
    {
      "index": 277,
      "start_seconds": 785,
      "source_timestamp": "13:05",
      "end_seconds": null,
      "text": "Okay. So we picked 100 random smart"
    },
    {
      "index": 278,
      "start_seconds": 788,
      "source_timestamp": "13:08",
      "end_seconds": null,
      "text": "contracts, smart contracts that have"
    },
    {
      "index": 279,
      "start_seconds": 790,
      "source_timestamp": "13:10",
      "end_seconds": null,
      "text": "been used on chain recently and uh we're"
    },
    {
      "index": 280,
      "start_seconds": 793,
      "source_timestamp": "13:13",
      "end_seconds": null,
      "text": "running these tools offline meaning that"
    },
    {
      "index": 281,
      "start_seconds": 795,
      "source_timestamp": "13:15",
      "end_seconds": null,
      "text": "you know storage is symbolic and and"
    },
    {
      "index": 282,
      "start_seconds": 798,
      "source_timestamp": "13:18",
      "end_seconds": null,
      "text": "that sort of thing."
    },
    {
      "index": 283,
      "start_seconds": 800,
      "source_timestamp": "13:20",
      "end_seconds": null,
      "text": "Now it turns out that a lot of tools"
    },
    {
      "index": 284,
      "start_seconds": 802,
      "source_timestamp": "13:22",
      "end_seconds": null,
      "text": "they are not geared towards exploration"
    },
    {
      "index": 285,
      "start_seconds": 805,
      "source_timestamp": "13:25",
      "end_seconds": null,
      "text": "meaning that they are geared towards you"
    },
    {
      "index": 286,
      "start_seconds": 807,
      "source_timestamp": "13:27",
      "end_seconds": null,
      "text": "know like um checking a single property"
    },
    {
      "index": 287,
      "start_seconds": 811,
      "source_timestamp": "13:31",
      "end_seconds": null,
      "text": "not at checking all the executions of of"
    },
    {
      "index": 288,
      "start_seconds": 813,
      "source_timestamp": "13:33",
      "end_seconds": null,
      "text": "code bases. And so in the case of Halmos"
    },
    {
      "index": 289,
      "start_seconds": 816,
      "source_timestamp": "13:36",
      "end_seconds": null,
      "text": "we wrote a custom harness and we needed"
    },
    {
      "index": 290,
      "start_seconds": 818,
      "source_timestamp": "13:38",
      "end_seconds": null,
      "text": "to fiddle a lot with the settings to to"
    },
    {
      "index": 291,
      "start_seconds": 821,
      "source_timestamp": "13:41",
      "end_seconds": null,
      "text": "make it run uh this experiment. All"
    },
    {
      "index": 292,
      "start_seconds": 824,
      "source_timestamp": "13:44",
      "end_seconds": null,
      "text": "tools use Z3 120 second timeout which is"
    },
    {
      "index": 293,
      "start_seconds": 828,
      "source_timestamp": "13:48",
      "end_seconds": null,
      "text": "like really low. Um and all the"
    },
    {
      "index": 294,
      "start_seconds": 831,
      "source_timestamp": "13:51",
      "end_seconds": null,
      "text": "exploration parameters you know it takes"
    },
    {
      "index": 295,
      "start_seconds": 833,
      "source_timestamp": "13:53",
      "end_seconds": null,
      "text": "a long time to configure these tools to"
    },
    {
      "index": 296,
      "start_seconds": 835,
      "source_timestamp": "13:55",
      "end_seconds": null,
      "text": "make sure that they're level playing"
    },
    {
      "index": 297,
      "start_seconds": 838,
      "source_timestamp": "13:58",
      "end_seconds": null,
      "text": "field that they're all exploring the"
    },
    {
      "index": 298,
      "start_seconds": 839,
      "source_timestamp": "13:59",
      "end_seconds": null,
      "text": "same number of paths and whatnot. Okay."
    },
    {
      "index": 299,
      "start_seconds": 843,
      "source_timestamp": "14:03",
      "end_seconds": null,
      "text": "So in the case of our tool we managed to"
    },
    {
      "index": 300,
      "start_seconds": 846,
      "source_timestamp": "14:06",
      "end_seconds": null,
      "text": "analyze 80% of the smart contracts all"
    },
    {
      "index": 301,
      "start_seconds": 850,
      "source_timestamp": "14:10",
      "end_seconds": null,
      "text": "possible parts. Okay."
    },
    {
      "index": 302,
      "start_seconds": 852,
      "source_timestamp": "14:12",
      "end_seconds": null,
      "text": "um except loops"
    },
    {
      "index": 303,
      "start_seconds": 855,
      "source_timestamp": "14:15",
      "end_seconds": null,
      "text": "in the case of HVM 31% and halmos only"
    },
    {
      "index": 304,
      "start_seconds": 859,
      "source_timestamp": "14:19",
      "end_seconds": null,
      "text": "26%. Okay, a lot of these tools have"
    },
    {
      "index": 305,
      "start_seconds": 862,
      "source_timestamp": "14:22",
      "end_seconds": null,
      "text": "partial results they actually most of"
    },
    {
      "index": 306,
      "start_seconds": 865,
      "source_timestamp": "14:25",
      "end_seconds": null,
      "text": "the time what they do is they choke for"
    },
    {
      "index": 307,
      "start_seconds": 867,
      "source_timestamp": "14:27",
      "end_seconds": null,
      "text": "instance when parsing call data or you"
    },
    {
      "index": 308,
      "start_seconds": 870,
      "source_timestamp": "14:30",
      "end_seconds": null,
      "text": "know moving things to memory that sort"
    },
    {
      "index": 309,
      "start_seconds": 872,
      "source_timestamp": "14:32",
      "end_seconds": null,
      "text": "of thing. So because of the loops."
    },
    {
      "index": 310,
      "start_seconds": 877,
      "source_timestamp": "14:37",
      "end_seconds": null,
      "text": "Okay. So uh we've uh deployed this tool"
    },
    {
      "index": 311,
      "start_seconds": 880,
      "source_timestamp": "14:40",
      "end_seconds": null,
      "text": "internally. We're using it for uh"
    },
    {
      "index": 312,
      "start_seconds": 883,
      "source_timestamp": "14:43",
      "end_seconds": null,
      "text": "auditing at DOB. Uh we also use it for"
    },
    {
      "index": 313,
      "start_seconds": 886,
      "source_timestamp": "14:46",
      "end_seconds": null,
      "text": "bug hunting. And so it's designed as"
    },
    {
      "index": 314,
      "start_seconds": 888,
      "source_timestamp": "14:48",
      "end_seconds": null,
      "text": "such. It's it's it's currently not"
    },
    {
      "index": 315,
      "start_seconds": 890,
      "source_timestamp": "14:50",
      "end_seconds": null,
      "text": "designed to uh to be a replacement of a"
    },
    {
      "index": 316,
      "start_seconds": 894,
      "source_timestamp": "14:54",
      "end_seconds": null,
      "text": "formal prover, but it could possibly"
    },
    {
      "index": 317,
      "start_seconds": 896,
      "source_timestamp": "14:56",
      "end_seconds": null,
      "text": "become one in the future. And right now"
    },
    {
      "index": 318,
      "start_seconds": 899,
      "source_timestamp": "14:59",
      "end_seconds": null,
      "text": "we're also doing explorations you know"
    },
    {
      "index": 319,
      "start_seconds": 901,
      "source_timestamp": "15:01",
      "end_seconds": null,
      "text": "trying to see whether um you know"
    },
    {
      "index": 320,
      "start_seconds": 903,
      "source_timestamp": "15:03",
      "end_seconds": null,
      "text": "certain EIPs will break contracts in the"
    },
    {
      "index": 321,
      "start_seconds": 906,
      "source_timestamp": "15:06",
      "end_seconds": null,
      "text": "future once they are deployed and and"
    },
    {
      "index": 322,
      "start_seconds": 909,
      "source_timestamp": "15:09",
      "end_seconds": null,
      "text": "that sort of stuff"
    },
    {
      "index": 323,
      "start_seconds": 911,
      "source_timestamp": "15:11",
      "end_seconds": null,
      "text": "allchain analysis kind of like what we"
    },
    {
      "index": 324,
      "start_seconds": 914,
      "source_timestamp": "15:14",
      "end_seconds": null,
      "text": "were doing with static analysis."
    },
    {
      "index": 325,
      "start_seconds": 916,
      "source_timestamp": "15:16",
      "end_seconds": null,
      "text": "Any questions?"
    },
    {
      "index": 326,
      "start_seconds": 924,
      "source_timestamp": "15:24",
      "end_seconds": null,
      "text": ">> Okay. Yep."
    },
    {
      "index": 327,
      "start_seconds": 928,
      "source_timestamp": "15:28",
      "end_seconds": null,
      "text": "just light on how you kind of cutting"
    },
    {
      "index": 328,
      "start_seconds": 930,
      "source_timestamp": "15:30",
      "end_seconds": null,
      "text": "the cost down of using agent workflows"
    },
    {
      "index": 329,
      "start_seconds": 933,
      "source_timestamp": "15:33",
      "end_seconds": null,
      "text": "in symbolic execution."
    },
    {
      "index": 330,
      "start_seconds": 935,
      "source_timestamp": "15:35",
      "end_seconds": null,
      "text": ">> Yeah."
    },
    {
      "index": 331,
      "start_seconds": 935,
      "source_timestamp": "15:35",
      "end_seconds": null,
      "text": ">> Hi, I'm Manu from lip P2P. Amazing talk."
    },
    {
      "index": 332,
      "start_seconds": 938,
      "source_timestamp": "15:38",
      "end_seconds": null,
      "text": "Just wish to ask how you are kind of"
    },
    {
      "index": 333,
      "start_seconds": 941,
      "source_timestamp": "15:41",
      "end_seconds": null,
      "text": "optimizing the cost of symbolic"
    },
    {
      "index": 334,
      "start_seconds": 944,
      "source_timestamp": "15:44",
      "end_seconds": null,
      "text": "execution when you're building a swarm"
    },
    {
      "index": 335,
      "start_seconds": 945,
      "source_timestamp": "15:45",
      "end_seconds": null,
      "text": "of agents and you're kind of using some"
    },
    {
      "index": 336,
      "start_seconds": 948,
      "source_timestamp": "15:48",
      "end_seconds": null,
      "text": "of the things to verify and also have"
    },
    {
      "index": 337,
      "start_seconds": 949,
      "source_timestamp": "15:49",
      "end_seconds": null,
      "text": "some of the agents to kind of do uh the"
    },
    {
      "index": 338,
      "start_seconds": 952,
      "source_timestamp": "15:52",
      "end_seconds": null,
      "text": "uh like the final job over here. Yeah."
    },
    {
      "index": 339,
      "start_seconds": 955,
      "source_timestamp": "15:55",
      "end_seconds": null,
      "text": "Thanks."
    },
    {
      "index": 340,
      "start_seconds": 955,
      "source_timestamp": "15:55",
      "end_seconds": null,
      "text": ">> Yeah. All right. So, first of all, you"
    },
    {
      "index": 341,
      "start_seconds": 958,
      "source_timestamp": "15:58",
      "end_seconds": null,
      "text": "got to optimize the um you know the"
    },
    {
      "index": 342,
      "start_seconds": 961,
      "source_timestamp": "16:01",
      "end_seconds": null,
      "text": "interface of the tool, make it"
    },
    {
      "index": 343,
      "start_seconds": 962,
      "source_timestamp": "16:02",
      "end_seconds": null,
      "text": "appropriate for command line um"
    },
    {
      "index": 344,
      "start_seconds": 965,
      "source_timestamp": "16:05",
      "end_seconds": null,
      "text": "environment and that sort of thing. The"
    },
    {
      "index": 345,
      "start_seconds": 967,
      "source_timestamp": "16:07",
      "end_seconds": null,
      "text": "output as I showed on one of the"
    },
    {
      "index": 346,
      "start_seconds": 970,
      "source_timestamp": "16:10",
      "end_seconds": null,
      "text": "previous slides is very very dense and"
    },
    {
      "index": 347,
      "start_seconds": 973,
      "source_timestamp": "16:13",
      "end_seconds": null,
      "text": "maps directly to the to the source"
    },
    {
      "index": 348,
      "start_seconds": 976,
      "source_timestamp": "16:16",
      "end_seconds": null,
      "text": "level. It's like compared to some other"
    },
    {
      "index": 349,
      "start_seconds": 979,
      "source_timestamp": "16:19",
      "end_seconds": null,
      "text": "tools out there. You know, other tools"
    },
    {
      "index": 350,
      "start_seconds": 981,
      "source_timestamp": "16:21",
      "end_seconds": null,
      "text": "will generate pages and pages and pages"
    },
    {
      "index": 351,
      "start_seconds": 983,
      "source_timestamp": "16:23",
      "end_seconds": null,
      "text": "of output. this one will generate you"
    },
    {
      "index": 352,
      "start_seconds": 986,
      "source_timestamp": "16:26",
      "end_seconds": null,
      "text": "know the very succinct things. So this"
    },
    {
      "index": 353,
      "start_seconds": 990,
      "source_timestamp": "16:30",
      "end_seconds": null,
      "text": "this is basically heavy compiler"
    },
    {
      "index": 354,
      "start_seconds": 992,
      "source_timestamp": "16:32",
      "end_seconds": null,
      "text": "engineering right it's compiler"
    },
    {
      "index": 355,
      "start_seconds": 994,
      "source_timestamp": "16:34",
      "end_seconds": null,
      "text": "engineering trying to map you know back"
    },
    {
      "index": 356,
      "start_seconds": 996,
      "source_timestamp": "16:36",
      "end_seconds": null,
      "text": "to source you know get get all the"
    },
    {
      "index": 357,
      "start_seconds": 998,
      "source_timestamp": "16:38",
      "end_seconds": null,
      "text": "debugging information through the"
    },
    {
      "index": 358,
      "start_seconds": 1000,
      "source_timestamp": "16:40",
      "end_seconds": null,
      "text": "compilation process it's that sort of"
    },
    {
      "index": 359,
      "start_seconds": 1002,
      "source_timestamp": "16:42",
      "end_seconds": null,
      "text": "work"
    },
    {
      "index": 360,
      "start_seconds": 1022,
      "source_timestamp": "17:02",
      "end_seconds": null,
      "text": "So my my question is what would you say"
    },
    {
      "index": 361,
      "start_seconds": 1026,
      "source_timestamp": "17:06",
      "end_seconds": null,
      "text": "is the greatest one of the major"
    },
    {
      "index": 362,
      "start_seconds": 1029,
      "source_timestamp": "17:09",
      "end_seconds": null,
      "text": "challenges in terms of adopting symbolic"
    },
    {
      "index": 363,
      "start_seconds": 1032,
      "source_timestamp": "17:12",
      "end_seconds": null,
      "text": "execution in the generic auditing"
    },
    {
      "index": 364,
      "start_seconds": 1035,
      "source_timestamp": "17:15",
      "end_seconds": null,
      "text": "ecosystem. Your form of verification you"
    },
    {
      "index": 365,
      "start_seconds": 1037,
      "source_timestamp": "17:17",
      "end_seconds": null,
      "text": "got fuzzing. you have so many different"
    },
    {
      "index": 366,
      "start_seconds": 1039,
      "source_timestamp": "17:19",
      "end_seconds": null,
      "text": "things is what what's the biggest"
    },
    {
      "index": 367,
      "start_seconds": 1041,
      "source_timestamp": "17:21",
      "end_seconds": null,
      "text": "challenge you would say? Well, the"
    },
    {
      "index": 368,
      "start_seconds": 1044,
      "source_timestamp": "17:24",
      "end_seconds": null,
      "text": "biggest challenge I thought, you know,"
    },
    {
      "index": 369,
      "start_seconds": 1046,
      "source_timestamp": "17:26",
      "end_seconds": null,
      "text": "when embarking on this project is that"
    },
    {
      "index": 370,
      "start_seconds": 1048,
      "source_timestamp": "17:28",
      "end_seconds": null,
      "text": "most tools don't work well enough to"
    },
    {
      "index": 371,
      "start_seconds": 1050,
      "source_timestamp": "17:30",
      "end_seconds": null,
      "text": "merit, you know, spending enough time um"
    },
    {
      "index": 372,
      "start_seconds": 1054,
      "source_timestamp": "17:34",
      "end_seconds": null,
      "text": "getting them up and running because most"
    },
    {
      "index": 373,
      "start_seconds": 1056,
      "source_timestamp": "17:36",
      "end_seconds": null,
      "text": "of the tools, you know, I I kind of like"
    },
    {
      "index": 374,
      "start_seconds": 1058,
      "source_timestamp": "17:38",
      "end_seconds": null,
      "text": "alluded to this. They're not designed to"
    },
    {
      "index": 375,
      "start_seconds": 1060,
      "source_timestamp": "17:40",
      "end_seconds": null,
      "text": "explore an entire contract. They're"
    },
    {
      "index": 376,
      "start_seconds": 1062,
      "source_timestamp": "17:42",
      "end_seconds": null,
      "text": "designed to look at a single function at"
    },
    {
      "index": 377,
      "start_seconds": 1064,
      "source_timestamp": "17:44",
      "end_seconds": null,
      "text": "a single property, you know, like a"
    },
    {
      "index": 378,
      "start_seconds": 1066,
      "source_timestamp": "17:46",
      "end_seconds": null,
      "text": "property on a single variable or"
    },
    {
      "index": 379,
      "start_seconds": 1068,
      "source_timestamp": "17:48",
      "end_seconds": null,
      "text": "something like that. So, that's very"
    },
    {
      "index": 380,
      "start_seconds": 1069,
      "source_timestamp": "17:49",
      "end_seconds": null,
      "text": "very time consuming."
    },
    {
      "index": 381,
      "start_seconds": 1072,
      "source_timestamp": "17:52",
      "end_seconds": null,
      "text": "um you know tools get better and better"
    },
    {
      "index": 382,
      "start_seconds": 1075,
      "source_timestamp": "17:55",
      "end_seconds": null,
      "text": "and then instead of using the tool you"
    },
    {
      "index": 383,
      "start_seconds": 1078,
      "source_timestamp": "17:58",
      "end_seconds": null,
      "text": "ask your AI agent to use the tool that's"
    },
    {
      "index": 384,
      "start_seconds": 1081,
      "source_timestamp": "18:01",
      "end_seconds": null,
      "text": "going to be free for you right you ask"
    },
    {
      "index": 385,
      "start_seconds": 1083,
      "source_timestamp": "18:03",
      "end_seconds": null,
      "text": "load hey use this tool on this and if it"
    },
    {
      "index": 386,
      "start_seconds": 1086,
      "source_timestamp": "18:06",
      "end_seconds": null,
      "text": "works you're going to get good results."
    },
    {
      "index": 387,
      "start_seconds": 1095,
      "source_timestamp": "18:15",
      "end_seconds": null,
      "text": "So basically when you reduce the the"
    },
    {
      "index": 388,
      "start_seconds": 1099,
      "source_timestamp": "18:19",
      "end_seconds": null,
      "text": "type to the integer that will be like"
    },
    {
      "index": 389,
      "start_seconds": 1101,
      "source_timestamp": "18:21",
      "end_seconds": null,
      "text": "more um more CV friendly type that could"
    },
    {
      "index": 390,
      "start_seconds": 1104,
      "source_timestamp": "18:24",
      "end_seconds": null,
      "text": "be quick make the solving time quicker"
    },
    {
      "index": 391,
      "start_seconds": 1106,
      "source_timestamp": "18:26",
      "end_seconds": null,
      "text": "basically like when you convert all"
    },
    {
      "index": 392,
      "start_seconds": 1108,
      "source_timestamp": "18:28",
      "end_seconds": null,
      "text": "these like type to be more solving"
    },
    {
      "index": 393,
      "start_seconds": 1110,
      "source_timestamp": "18:30",
      "end_seconds": null,
      "text": "friendly um for the SMT solver then"
    },
    {
      "index": 394,
      "start_seconds": 1113,
      "source_timestamp": "18:33",
      "end_seconds": null,
      "text": "what's the basically downside cost like"
    },
    {
      "index": 395,
      "start_seconds": 1115,
      "source_timestamp": "18:35",
      "end_seconds": null,
      "text": "when you face all this like BIP type and"
    },
    {
      "index": 396,
      "start_seconds": 1118,
      "source_timestamp": "18:38",
      "end_seconds": null,
      "text": "um potentially that could be like"
    },
    {
      "index": 397,
      "start_seconds": 1119,
      "source_timestamp": "18:39",
      "end_seconds": null,
      "text": "forcing the bit factor in the SMT lib"
    },
    {
      "index": 398,
      "start_seconds": 1122,
      "source_timestamp": "18:42",
      "end_seconds": null,
      "text": "language instead of forcing it to be"
    },
    {
      "index": 399,
      "start_seconds": 1124,
      "source_timestamp": "18:44",
      "end_seconds": null,
      "text": "integer that definitely faster but like"
    },
    {
      "index": 400,
      "start_seconds": 1126,
      "source_timestamp": "18:46",
      "end_seconds": null,
      "text": "what's the downside cost of all this"
    },
    {
      "index": 401,
      "start_seconds": 1128,
      "source_timestamp": "18:48",
      "end_seconds": null,
      "text": "address type all this dynamic factor"
    },
    {
      "index": 402,
      "start_seconds": 1130,
      "source_timestamp": "18:50",
      "end_seconds": null,
      "text": "type or dynamic array in the solidity"
    },
    {
      "index": 403,
      "start_seconds": 1133,
      "source_timestamp": "18:53",
      "end_seconds": null,
      "text": "because that must be some downside costs"
    },
    {
      "index": 404,
      "start_seconds": 1135,
      "source_timestamp": "18:55",
      "end_seconds": null,
      "text": "there yeah"
    },
    {
      "index": 405,
      "start_seconds": 1136,
      "source_timestamp": "18:56",
      "end_seconds": null,
      "text": ">> yeah there's there's lots of downsides I"
    },
    {
      "index": 406,
      "start_seconds": 1138,
      "source_timestamp": "18:58",
      "end_seconds": null,
      "text": "mean first of all for instance when when"
    },
    {
      "index": 407,
      "start_seconds": 1140,
      "source_timestamp": "19:00",
      "end_seconds": null,
      "text": "you treat a U in 256 as an idealized"
    },
    {
      "index": 408,
      "start_seconds": 1143,
      "source_timestamp": "19:03",
      "end_seconds": null,
      "text": "integer well it's not going to wrap"
    },
    {
      "index": 409,
      "start_seconds": 1145,
      "source_timestamp": "19:05",
      "end_seconds": null,
      "text": "around but if you know that you know"
    },
    {
      "index": 410,
      "start_seconds": 1148,
      "source_timestamp": "19:08",
      "end_seconds": null,
      "text": "that code which which it is in is an"
    },
    {
      "index": 411,
      "start_seconds": 1150,
      "source_timestamp": "19:10",
      "end_seconds": null,
      "text": "checked code all right then you can"
    },
    {
      "index": 412,
      "start_seconds": 1154,
      "source_timestamp": "19:14",
      "end_seconds": null,
      "text": "still treat it as an idealized integer"
    },
    {
      "index": 413,
      "start_seconds": 1156,
      "source_timestamp": "19:16",
      "end_seconds": null,
      "text": "and then you just add an additional"
    },
    {
      "index": 414,
      "start_seconds": 1158,
      "source_timestamp": "19:18",
      "end_seconds": null,
      "text": "check that it doesn't exceed 2 to the"
    },
    {
      "index": 415,
      "start_seconds": 1162,
      "source_timestamp": "19:22",
      "end_seconds": null,
      "text": "255"
    },
    {
      "index": 416,
      "start_seconds": 1163,
      "source_timestamp": "19:23",
      "end_seconds": null,
      "text": "26 minus one and it doesn't go below"
    },
    {
      "index": 417,
      "start_seconds": 1166,
      "source_timestamp": "19:26",
      "end_seconds": null,
      "text": "zero for instance. Okay, so you can use"
    },
    {
      "index": 418,
      "start_seconds": 1168,
      "source_timestamp": "19:28",
      "end_seconds": null,
      "text": "some of that type information and then"
    },
    {
      "index": 419,
      "start_seconds": 1170,
      "source_timestamp": "19:30",
      "end_seconds": null,
      "text": "it's just as sound as using bit vectors"
    },
    {
      "index": 420,
      "start_seconds": 1174,
      "source_timestamp": "19:34",
      "end_seconds": null,
      "text": "except that bit vectors which would be"
    },
    {
      "index": 421,
      "start_seconds": 1177,
      "source_timestamp": "19:37",
      "end_seconds": null,
      "text": "sound in this case uh if you try"
    },
    {
      "index": 422,
      "start_seconds": 1179,
      "source_timestamp": "19:39",
      "end_seconds": null,
      "text": "multiplication and division you're going"
    },
    {
      "index": 423,
      "start_seconds": 1181,
      "source_timestamp": "19:41",
      "end_seconds": null,
      "text": "to break the solvers because it's going"
    },
    {
      "index": 424,
      "start_seconds": 1183,
      "source_timestamp": "19:43",
      "end_seconds": null,
      "text": "to create circuits that are too big and"
    },
    {
      "index": 425,
      "start_seconds": 1186,
      "source_timestamp": "19:46",
      "end_seconds": null,
      "text": "there's a lot of tricks like that and"
    },
    {
      "index": 426,
      "start_seconds": 1188,
      "source_timestamp": "19:48",
      "end_seconds": null,
      "text": "frankly you do end up with some"
    },
    {
      "index": 427,
      "start_seconds": 1190,
      "source_timestamp": "19:50",
      "end_seconds": null,
      "text": "unsoundness if any of these algorithms"
    },
    {
      "index": 428,
      "start_seconds": 1193,
      "source_timestamp": "19:53",
      "end_seconds": null,
      "text": "are not perfect."
    }
  ]
}
