# Two Humans, One Table Each: Calling Israel's Election My Weird Prompts — an AI-generated podcast. Daniel Rosehill sends in a prompt; AI hosts Corn and Herman discuss it. Verify important claims independently. - Episode: 6204 - Published: 2026-10-22 - Duration: 25:51 - Audio: https://episodes.myweirdprompts.com/audio/israel-election-seat-by-seat-predictions-56192.m4a - Tags: israel, elections, forecasting-and-wargaming, haredi-community, israel-defense-forces - Web page: https://myweirdprompts.com/episode/israel-election-seat-by-seat-predictions-56192/ - JSON: https://myweirdprompts.com/api/episodes/israel-election-seat-by-seat-predictions-56192.json ## Summary Five days before the vote, with polls putting Netanyahu's bloc anywhere from the mid-forties to sixty-two depending on who you read, this episode throws out the forecasting models and does it by hand. Two hosts build complete 120-seat tables independently — Corn nudging the mainstream average, Herman betting the pollsters are missing the same voters they missed in 2022 — and then argue over every seat that separates them. Along the way: why Shas is the hinge, whether new lists like Yashar get over-polled, what happens when the largest list and the most able coalition are two different things, and how the prediction markets price it all. ## The prompt > Israel Votes, Part 7: five days before the 27 October 2026 Knesset election, and after the Part 6 decision-model experiment, Daniel joins Herman and Corn in the studio and asks for their own predictions from the current news. Claude set two rival seat tables: Corn argues a conservative call (polls roughly right, Yashar largest, Netanyahu bloc 50) and Herman a volatile one (historical poll misses and threshold casualties, Likud largest, Netanyahu bloc 59). Daniel doesn't give his own prediction; he questions them. ## Show notes Five days out from the election, the polls disagree with each other by more than ten seats on Netanyahu's bloc, so this episode sets the models aside and does it the old way: two complete 120-seat tables, written separately, argued out loud. Corn's table is the conservative one — the mainstream average is roughly right, Yashar at twenty-two, Likud nudged up two seats to twenty-one, and Netanyahu's bloc landing at fifty. Herman's is the contrarian one — Likud at twenty-four, Shas surging to ten, Amcha Yisrael clearing the threshold, and the bloc at fifty-nine, two seats short of a majority but enough for the first mandate. The argument underneath is about polling error. Herman points to 2022, when the mainstream final polls undercounted the bloc by three to five seats, and to Shas specifically, which has consistently outperformed its polling. Corn counters that one miss in three elections isn't a law of nature, and that Herman's table requires four separate things to break the same way. Both agree on the shape of what comes next: nobody reaches sixty-one, the first week is a quiet shouting match, and the reported Likud effort to lure Yashar lawmakers is exactly the kind of two-seat gap that gets hunted. The markets sit between the two tables — Eisenkot around forty-nine cents to be prime minister, Netanyahu at thirty-one. ## Transcript Corn: Every mainstream poll this week puts Netanyahu's bloc somewhere between the mid forties and fifty-one, and the headlines read that as the opposition winning. It isn't winning anything. The prize is sixty-one seats, and nobody in the mainstream gets within hearing distance of it. Herman: Channel Fourteen has Netanyahu's bloc at sixty-two, so the headline depends on which pollster you read over breakfast. Five days out, and the pollsters differ on the bloc by more than ten seats. Corn: So this week we're doing it differently. Daniel is in the studio, and one note on that, said plainly once: his lines are voiced by an AI clone of his voice. Also, this script was written offline, by hand, rather than by the usual pipeline model. This is Part Seven of the election series. Daniel: Part Six was two decision models and a pile of frozen evidence. This time I want you two. No models, no forward passes. Read the news as it stands today, the twenty-second, and tell me what you think happens on Sunday. I'm not giving my own call. I'm going to ask you things, and I want numbers. Herman: Seat by seat? Daniel: Seat by seat. A full table each, adding up to a hundred and twenty. Corn: Then let's be clear about what these are. Two animals' opinions, formed from the news. Not a forecast from a model, and nobody should vote by either of them. Herman: We also wrote our tables separately. Corn hasn't seen mine and I haven't seen his, so whatever friction you hear is real friction. Corn: I'll go first, because mine is the dull one. My call is that the polls are roughly right. Yashar, Eisenkot's list, twenty-two. Likud twenty-one. Together, the Bennett and Lapid merger, thirteen. Daniel: Hang on. The seven latest mainstream polls average Likud at around nineteen. You've got twenty-one. Corn: I have, and I own it. Yashar averages about twenty-two across those seven, so I left it alone. Likud averages nineteen and a bit, and I gave it two seats back, because final polls have leaned against Netanyahu's side before and I'm not a fanatic. That's the whole of my adjustment. A nudge. Herman: Two seats is a nudge for a sloth. For anyone else it's a lurch in slow motion. Corn: Everything I do is in slow motion. It's why you can see it coming. Daniel: Go on with the middle of the table. Corn: The Democrats nine. The Joint List nine. Yisrael Beiteinu eight. Daniel: Keep going. Corn: Then Otzma Yehudit eight, Shas eight, United Torah Judaism seven. Herman: Shas at eight. I'll come back to that one. Corn: Religious Zionism with Zehut six, Ra'am five, and the Reservists, Hendel and Zelekha's list, four. Amcha Yisrael misses the threshold, around three percent. That comes to a hundred and twenty. Daniel: What does that table mean for a government? Corn: Netanyahu's bloc, Likud plus Shas plus United Torah Judaism plus Otzma plus Religious Zionism, comes to fifty. The Zionist opposition, Yashar and Together and the Democrats and Yisrael Beiteinu and the Reservists, is fifty-six. The two Arab lists have fourteen between them. Daniel: So nobody gets to sixty-one. Corn: Nobody. Yashar is the largest list, and Eisenkot most likely gets the first mandate. After that it's a long, ugly crawl. Fifty-six plus Ra'am's five is exactly sixty-one, which is the twenty twenty-one government again with nothing to spare. Or there's some unity arrangement with Likud. Either way it's a crawl, and a crawl is the one part of this I'm qualified for. Daniel: Why trust the polls at all? They've been wrong before. Corn: Because the mainstream average has been steady, as far as I can see, and Gantz quitting on the nineteenth is already in it. And the lesson people draw from past misses is overfitted. In twenty twenty-one the final polls had Netanyahu's side between forty-nine and fifty-three, and the result was fifty-two. Dead in the range. One miss in the last three isn't a law of nature. Herman: It's the most recent miss, though, so I'd read it as the pattern, not the exception. My call is that the error is the story. Likud twenty-four. Yashar nineteen. Together twelve. Daniel: Likud largest. Herman: By five seats. Then Shas ten, the Democrats nine, United Torah Judaism eight. Corn: Shas at ten. Brave. Herman: Yisrael Beiteinu eight, Otzma eight, the Joint List eight. Daniel: Nearly there. Herman: Religious Zionism with Zehut five, Ra'am five, Amcha Yisrael four, and the Reservists zero. That's a hundred and twenty. I checked it twice. Corn: He checks everything twice. It's why dinner is late. Herman: Netanyahu's bloc is fifty-nine, and I'll be exact, because that includes Amcha's four. Without them it's fifty-five. The opposition is forty-eight and the Arab lists thirteen. So he's two seats short of sixty-one, he gets the first mandate as the larger list, and he goes hunting. Daniel: Hunting for what? Herman: Defectors. Likud has reportedly set up a team to lure Yashar lawmakers after the vote, and Yashar says its slate holds. I'd call that a report, not a fact. But two seats is exactly the kind of gap a team like that is built for. Corn: Your whole government is two people changing their minds. Herman: Which is how most governments here end, so I'm not embarrassed. Daniel: Herman, justify it. Seven mainstream polls have Likud behind Yashar. Where does a five-seat Likud lead come from? Herman: Three places. First, in twenty twenty-two the mainstream final polls had Netanyahu's bloc at fifty-nine to sixty-one, and it got sixty-four. Three to five seats under. Corn: That's one election. Herman: Second, the under-count lives in Shas. In twenty twenty-two Shas polled eight to nine and got eleven. In twenty nineteen, four to six, and it got eight. Daniel: Go on. Herman: Third, new lists get over-polled. New Hope in twenty twenty-one polled seven to eleven and got six. Yamina polled eight to twelve and got seven. Yashar is a new list, so I take seats off it and hand them to the old machines. Daniel: But Likud itself was inside its range in twenty twenty-two. The miss was Shas and the religious right, not Likud. Herman: Correct, which is why only part of my bloc is Likud. In twenty nineteen, though, Likud polled twenty-six to thirty-one, and Globes had the final result at thirty-six. Daniel: Is that all of it? Herman: No. And then there's Channel Fourteen, at thirty for Likud today. In twenty twenty-two its pollster had the bloc at sixty-two, two under the result, while the mainstream was three to five under. Corn: You're half-endorsing Channel Fourteen. Herman: I'm not. Thirty is nowhere I'd go, and they could easily be wrong this time. But the direction of the old error was right, so I walked a quarter of the way toward them. If you've seen higher Likud averages elsewhere, that's usually why. They have Channel Fourteen and Direct Polls mixed in. Corn: A quarter of the way is a lot for a donkey who calls himself a numbers man. Herman: Corn, you moved Likud two seats on a feeling and called it a nudge. You can't hold a nudge against me. Daniel: Press each other, then. Corn, where's the weakest point in his table? Corn: The bloc. Fifty-nine means everything breaks one way. Shas surges, Amcha clears the line, the Reservists miss, Yashar slips. That's four coins landing on heads. Each one is plausible, and the four together is a tall order. Herman: The coins aren't independent, though, and that's my whole case. If the pollsters are all missing the same voters, and in twenty twenty-two they were, the errors arrive together. Four coins stuck to one magnet. Daniel: Herman, the weakest point in Corn's? Herman: That nothing in it surprises anybody. Every election in this series has had a surprise, and a table that treats the poll average as the truth has no room for one. A conservative call is also a bet that the pollsters have finally fixed it. Corn, do you actually believe that? Corn: I believe they've fixed enough of it. A smaller miss than twenty twenty-two is plenty for my table to survive. Daniel: And are the polls finally stable? Is that what you're betting? Herman: Before Corn answers, a voicemail, from Franklin in Dublin. Franklin: Hi lads, Franklin here, calling from the back of the forty two bus stuck in traffic on the Rathgar Road in Dublin, so you have me for at least another ten minutes. I wanted to say I loved the Israel Votes series, all seven parts, and I think your seat predictions are grand, though I'd say the answer is thirty nine seats for the biggest party and the rest is guesswork dressed up as maths. My question is this, and I know it's not really about Israel, but my brother in law runs a small bakery in Cork and he swears the bread rises better on the day the polls close. Is there any chance that's real, or is he just a man who likes an excuse to sit down? Anyway, keep going, I'm with you till the ads. Corn: Thanks, Franklin. Your brother in law has found himself a good excuse and I respect that. Back to Daniel's question, yes, that's what I'm betting. Stable doesn't mean the polls are right, only that the average has stopped moving, and I think the average is close enough. Herman: Stable at the average, I'll grant you. Underneath it's a different story. Yashar across the seven latest runs from twenty to twenty-three. Likud from eighteen to twenty-one. Those are the parts the average is built from. Daniel: So the average is steady and the polls aren't. Herman: A calm surface over a lot of moving parts. How much of the calm is real is exactly what we're guessing at. Daniel: Suppose Likud is the largest list in the end, but Eisenkot's camp has more seats than Netanyahu's. Who gets the call from the president? Corn: Whoever can show the president a path to sixty-one, and that's a count of recommendations, not of list sizes. On my table that's Eisenkot, but only if Liberman and Ra'am can both be named in the same breath, and I'm not sure they can. It's possible Eisenkot gets the mandate and the paperwork doesn't close. Herman: On mine the largest list wins the call by default, because nothing else gets near sixty-one. The trouble is that largest and able are two different things. Netanyahu at fifty-nine has to turn two seats from a rival's list into votes, and the rival's whole pitch is that they won't move. Corn: Which makes the first week after the vote a very quiet shouting match. Herman: And a lot of telephone calls. I'd be surprised if either camp's lists hold together for the full period. Nobody has ever described these as watertight. Daniel: Corn, does your table have a version where Netanyahu ends up prime minister anyway? Corn: Yes, the unity door. Likud twenty-one, Yashar twenty-two, and neither side reaching sixty-one is exactly the arithmetic that makes people talk about a unity government. The market says Eisenkot, about forty-nine cents. I'd give Netanyahu a bit more than the thirty-one cents it gives him, but not much more. I don't see him with more than fifty seats. Herman: And mine has Eisenkot as prime minister only if the defector hunt fails. I would put that at roughly a coin flip, which is a very different statement from saying I expect Eisenkot to win. Daniel: Where do the markets sit? Corn: Between us, mostly. The last snapshot had Eisenkot at about forty-nine cents to be prime minister, Netanyahu at thirty-one, Bennett at sixteen. Herman: Yashar as the largest list was seventy cents. So on who's biggest, the market sides with Corn. On the prime minister it's less sure than either of us, and Netanyahu at thirty-one cents is more than I'd have guessed for someone who in my table gets the first mandate. Corn: Their seat medians are in between too. Likud nearly twenty-two, Yashar a bit over twenty-two. Neither of our tables is where they are. Daniel: Is a price a forecast? Corn: It's a crowd of opinions with money behind them, which is more than either of us has. Herman: I'd take that as a promise that you'll bet on yours. Corn: I'd take it as a promise to bet slowly. Daniel: Go to the bottom of the table. The threshold. Herman: This is where my call lives. Three point two five percent is about four seats, and several lists are sitting on it. Take Amcha Yisrael, Ofer Winter's list. Midgam on the twenty-first had it at four seats. Daniel: And the others? Herman: Channel Thirteen the day before had it at three point two percent, out by a hair. Walla on the nineteenth had two point nine. Kan had it out. Corn: And the Reservists do the same trick in reverse. Herman: Mirror image, in the same week. Midgam has the Reservists out, at two point eight percent. Channel Thirteen, Walla and Kan all have them at four seats. Both lists flip between zero and four, and with them on or near Netanyahu's side of the ledger, that's a swing of a couple of seats on the bloc from nothing but a decimal. Corn: So I put the Reservists in and Amcha out, because three of the four polls say so. Majority rule. Herman: The Reservists at zero is also the number I'd bet is wrong. I'm the opposite of you on both. I have the history, though, and none of it is cheerful. New Right missed in twenty nineteen with three point two two percent. Meretz in twenty twenty-two with three point one six. Those aren't polling errors. A hair on the wrong side of a line. Daniel: What about Gantz? He quit on the nineteenth. Herman: Blue and White is still on the ballot. It polled somewhere between about half a percent and two percent before he withdrew. It gets no seats whatever happens. What's interesting is where those voters go, and some of them won't have read the news. Corn: I gave it no seats and no thought, which is roughly its poll standing. Daniel: Does the withdrawal help either of you? Corn: It's priced in. People read it Monday, pollsters fielded Tuesday and Wednesday, and nothing visible moved. It took away a list that wasn't winning a seat. Herman: There's also a report that coalition members have been urging small right-wing lists near the line to follow Gantz out. That's one snippet and I can't pin it down. If one of them did, it would move four seats in a single direction. Worth watching. Daniel: The Haredi parties. Corn, Shas is eight in yours. Herman has it at ten. Herman: The most consistent error in the whole history. Shas was under-polled in twenty nineteen and again in twenty twenty-two, by something like two to four seats each time. And this year the draft is on the ballot. Deri has said Shas won't join a government without a legislated exemption, and United Torah Judaism's Asher says the same, reportedly. The High Court annulled the temporary enforcement halt in September. That's a community with a very direct reason to turn out. Corn: I'd concede the turnout argument and keep my number. Shas at eight is already above the seven and a half in the polls. Pollsters miss a Haredi voter for the same reason I miss a morning. They aren't where you expect. Herman: So you concede the fact and keep the opinion. Corn: Yes. It's the brotherly arrangement. Daniel: And the Arab vote? The Joint List is back together, Hadash and Ta'al and Balad. Corn: Nine for me, with Ra'am five on its own. The polls have the Joint List at eight or nine. Herman: Eight for me. In twenty twenty-one the Joint List polled seven to ten and got six, so I shave. Turnout decides it. A Dayan Center survey in May reportedly projected fifty-three percent Arab turnout, up to sixty-seven if the Joint List reformed, and I haven't seen anything confirm it since. Overall turnout in twenty twenty-two was a little over seventy percent. Corn: So a revived Joint List is either a surge or the rumour of one. Daniel: How much does it matter that Ra'am runs separately? Corn: To my table it matters most, because fifty-six plus Ra'am's five is how Eisenkot gets to sixty-one. And Liberman has already called Eisenkot a second Gantz and accused him of planning a unity government with Likud. Whether Liberman sits in a coalition that leans on Ra'am, I can't tell you. Herman: I suspect Liberman can't either. Daniel: Anything in the late news that changes either table? Herman: Gantz is the big one, and we've done that. Reports today say Iran struck four vessels in the Gulf and warned the United States of a broader response. I don't know that it moves a single Israeli voter. It makes the incumbent's security argument louder, and that argument isn't new. Corn: Then there's the IDF assets story. An election official reportedly said that a Netanyahu party used IDF assets in its campaign ads illegally. I've seen that in one place only and I wouldn't price it. Herman: Both camps also reportedly expect Netanyahu to try squeezing Ben-Gvir's list toward the line. Otzma is at eight in both of our tables and seven to nine in the polls, so it's nowhere near, but the story is out there. Corn: Five days left, and nothing on that list has the weight to move more than a seat or two. Herman: A list at the line is a bigger risk than the Gulf, as far as my table is concerned. Hilbert: You're both right. Mostly. My Aunt Bernadette runs the sweepstake on this at Tuckett's, the chiropodist on Brent Street. Two pounds forty a card, written on the back of an appointment card. Thirty-one cards in. Every table has to add up to a hundred and twenty and she checks it with a wooden ruler. The ruler does nothing for the arithmetic. She likes the sound it makes against the desk. Corn: A ruler at least can't overfit. Hilbert: Tuesday she threw out mine. A hundred and nineteen and a half seats. I gave half a seat to Shas. She says the Knesset does not issue halves. Mr Tuckett handed in a hundred and twenty and a half and she allowed it, because the half was for his own party. He has entered a party. It is called the Tuckett Alliance and it polls three point one percent in the waiting room. She rounded it down to zero. Her threshold is four point two. She set that in the spring, at a meeting, with the door shut. Herman: Four point two would take out most of the small lists we've been through, Amcha and the Reservists included. She runs a harder Knesset than the real one. Hilbert: Herman, you said your sum out loud, to check it. Nobody at Tuckett's does that. The pot is held by Miss Odell, the receptionist, who is also the returning officer. She also runs the other sweepstake, the one on the second floor above the dentist, two pounds sixty, biscuit included. They do not speak. They keep their cards in separate drawers, and on Thursday six of my aunt's turned up in the dentist's. Nobody has explained how. The drawer was locked and the key is in a mug with a picture of a lighthouse on it. Corn: Six out of thirty-one is a fifth of the vote, give or take. Hilbert: She noticed. She sent the cards in for certification. All thirty-one, to the Central Elections Committee in Jerusalem, second class. She got the address off a tea towel. There has been no reply and she has taken that as certification. She has informed Miss Odell that the cards are now the property of the State of Israel and that Miss Odell may collect them from there. Miss Odell has asked for the postcode. Herman: Certification by silence is not how the real one works. That comes with a result and a date, and this year it's around the fifth of November. Hilbert: The fish tank has a vote. It has been empty since the council order, and she says that gives it a stake. It voted for the Tuckett Alliance. Miss Odell ruled it spoiled because the tank was not present. My aunt says the tank has not missed a day since they put in the lino. It is the only ballot the Alliance has received that was not handed in by Mr Tuckett. Corn: Best attendance record in the building, then. Hilbert: Three biscuits, counted out in front of it. Mr Tuckett has asked what they are for. Nobody has told him. Thursday he read the interim figures to a patient with his shoes off, and the patient voted, which is not allowed, because the patient was only there for a verruca. Corn, you say a number slowly until it sounds like a decision. She does the same. Herman: Does anyone actually win? Hilbert: Last time she paid out on the twenty-third day. In stamps. Corn: Twenty-three days is optimistic for what comes after the count here. Daniel: Right. Before we stop, I want each of you to name the one number in your table you'd most bet is wrong. Corn: Likud twenty-one. It's where I leaned away from the polls, so it's where I'm most exposed. If the polls are right, it's two seats high. If Herman is right, it's three low. The only number I'd defend with a shrug. Herman: The Reservists at zero. Three of the four polls with them have them at four. If they clear the line, my opposition goes to fifty-two and my bloc story shrinks, because the extra four seats have to come out of somewhere. So the number my argument leans on is the one I trust least. Corn: That's a very good thing to say about your own table. Herman: I'm a retired pediatrician. I'm used to being told I'm wrong by people who are three years old. Daniel: How will you score this after the results? Part Six had Brier scores for Clef and Jev. Herman: Certification is around the fifth of November, nine days after the vote. The seat table is a point call, not a probability, so Brier doesn't apply to it directly. We add up the absolute miss on every list. A perfect table scores zero, and the worst possible is two hundred and forty. Corn: Brier for the yes or no calls, then. We each commit to a percentage now. Mine is that Yashar is the largest list, at seventy percent, which happens to be the market. And that nobody reaches sixty-one, at sixty-five. Herman: Likud largest, at fifty-five percent. Nobody at sixty-one, fifty. And that Netanyahu gets the first mandate, at sixty. Daniel: If you're both wrong? Corn: Then the market was wrong first, and I'll blame them. Daniel: Give me the two calls side by side before we stop. Corn first. Corn: Yashar twenty-two, Likud twenty-one, Together thirteen. Netanyahu's bloc at fifty, the Zionist opposition at fifty-six, and a long ugly crawl for Eisenkot. Herman: Likud twenty-four, Yashar nineteen, Together twelve. Netanyahu's bloc at fifty-nine, the opposition at forty-eight, and Netanyahu two short of sixty-one with a defector team on the phone. Daniel: And neither of you is claiming a model. Corn: A sloth, a donkey and a pile of news. We said it at the start and I'm saying it again at the end. Herman: Whatever happens on Sunday, a few lists on either side of three point two five percent decide which of our tables looks clever. Corn: The count is nine days off, and after it comes the part nobody has put a price on, which is how long it takes before anyone gets paid. That's the stretch of the crawl I'd watch. Herman: Thanks to our producer, Hilbert Flumingtop, for keeping the show on the air. Corn: This has been My Weird Prompts. You can find every episode at my weird prompts dot com. Herman: See you tomorrow.