How many conflict deaths will there be in Sudan in 2025?
Phil’s pgodzinai is a side competitor with the machine forecasting bots of the VIEWS competition.
Perplexity response:
{think}
Okay, so I need to figure out how many state-based conflict deaths ACLED will report in Sudan for 2025. Let me start by understanding the question properly. The user wants to know the number of state-based (which I assume refers to conflicts involving government forces) deaths as recorded by ACLED this year. The options are from less than 1,000 up to more than 12,000.
First, I should check the provided search results for any data on ACLED’s recent reports. Looking at result [1], it mentions ACLED reporting over 28,700 deaths by late 2024. That’s a key point. Then result [2] from Arab News on May 5, 2025, also cites ACLED’s death toll as 28,700+ up to that date. Wait, but that’s up to May 2025? Let me check the date on result [2]. The URL says 2025-05-05, so that’s May 5th, 2025. So as of early May 2025, ACLED’s figure is already over 28,700. Wait, but the user is asking about the total for the entire year 2025. If by May 5th, the toll is 28,700+, then the annual number could be higher. But that seems conflicting because the previous year’s total was 28,700 by late 2024. Wait, maybe there’s a confusion here.
Wait, result [1] says that ACLED reported over 28,700 deaths by late 2024, and result [2] as of May 5, 2025, also mentions 28,700+ death toll. That doesn’t make sense unless the conflict continued into 2025. Wait, maybe the user is asking for 2025 data, but the search results are from early 2025. So perhaps the 28,700 is up to April or May 2025, not cumulative from previous years. Let me read the exact wording. Result [1] says “Estimates of fatalities vary widely, with ACLED reporting over 28,700 deaths by late 2024, while Genocide Watch suggests numbers as high as 150,000 when …”. So ACLED’s 28,700 is up to late 2024. Then result [2] from May 2025 says “28,700+ Conflict’s death toll as per ACLED records.” So as of May 2025, the death toll is over 28,700. If that’s the case, then the total for 2025 would be the difference between the current number and the previous year’s total. But wait, the user is asking for the number reported by ACLED in 2025. If the conflict continued into 2025, then ACLED would report deaths from January 1 to December 31, 2025. However, the search results don’t have data beyond May 2025, so we have to extrapolate.
Another thing to consider is the rate of deaths. Result [2] mentions that between January 31 and February 5, 2025, the UN reported 275 civilian deaths due to shelling and airstrikes. That’s 275 deaths over 5 days, which averages about 55 deaths per day. Extrapolating that to a year would be 55 * 365 ≈ 20,000, but this is a rough estimate and might not account for varying intensities of conflict.
Result [4] from May 3, 2025, talks about increasing drone attacks by RSF, with recent incidents causing 12 deaths. ACLED data shows RSF increasing drone attacks after losing control of Khartoum in March. So the conflict is still ongoing, and the death toll is rising. If the current toll is already over 28,700 by May 2025, and assuming the conflict continues at similar or increased rates, the total for 2025 could be higher. But how much higher?
Result [3] mentions global political violence increasing by 25% in 2024, with 223,000 people killed. While this is global data, it indicates a trend of increasing violence, which might suggest similar or worse figures in Sudan for 2025. However, Sudan’s specific data isn’t directly covered here.
Result [5] seems incomplete, so I’ll ignore it.
Now, looking at the options, the highest is “More than 12,000”. If the death toll as of May 2025 is already over 28,700, that would imply that the total for 2025 (assuming the year started in January) would be part of that 28,700. Wait, but that doesn’t make sense because the 28,700 mentioned in result [2] is up to May 2025, which would include deaths from January to May. So the 28,700 is up to May 5, 2025. If that’s the case, then the total for the entire year 2025 could be much higher. However, this might be a misunderstanding. Let me check the dates again.
Result [1] is dated May 2, 2025, and states ACLED reported over 28,700 deaths by late 2024. Result [2] from May 5, 2025, says the death toll is 28,700+ as per ACLED. So perhaps the 28,700 figure is cumulative from the start of the conflict in April 2023 up to May 2025. That would mean that in 2024, the toll was 28,700, and in 2025, it’s adding to that. But the user is asking for the number in 2025. If the conflict has been ongoing since 2023, the 28,700 might include previous years. This is a bit confusing. Alternatively, maybe the 28,700 is the total up to May 2025, meaning that the entire year’s toll (from January to May) is 28,700, which would suggest a very high monthly rate. For example, 28,700 over 5 months is about 5,740 per month, which would project to 68,880 for the year. But that seems extremely high and likely incorrect.
Wait, perhaps there’s a mistake in interpreting the data. Let me clarify:
Result [1] (May 2, 2025): “Estimates of fatalities vary widely, with ACLED reporting over 28,700 deaths by late 2024, while Genocide Watch suggests numbers as high as 150,000 when …”
Result [2] (May 5, 2025): “28,700+ Conflict’s death toll as per ACLED records.”
So ACLED’s figure as of May 5, 2025, is 28,700+, but this is cumulative from the start of the conflict in April 2023. Therefore, the total deaths from April 2023 to May 2025 are over 28,700. The user is asking for the number reported in 2025. So we need to know how many deaths occurred from January 1, 2025, to December 31, 2025, according to ACLED.
But the search results don’t provide a monthly breakdown. Result [2] mentions that between January 31 and February 5, 2025, 275 civilians were killed. That’s a 5-day period. If we take that as a rate, 275 deaths in 5 days is 55 per day. Over a year, that’s 55 * 365 ≈ 20,000. But this is just an extrapolation and might not account for varying conflict intensities.
Additionally, result [4] from May 3, 2025, notes that ACLED data shows RSF increasing drone attacks after losing control of Khartoum in March, leading to more strikes. This suggests an escalation, which could mean higher death tolls in the latter part of 2025.
Given that the total up to May 2025 is over 28,700, and assuming the conflict continues at a similar or higher rate, the annual toll for 2025 could be in the higher ranges. However, without specific monthly data for 2025, it’s challenging. The UN data from result [2] shows 275 deaths in 5 days in February, which is a high rate. If that rate continued, the annual toll would be around 20,000, but that seems too high. Alternatively, if the majority of the 28,700 deaths occurred in 2024, then 2025’s number might be lower. But the search results don’t provide a clear split between 2024 and 2025.
Another angle: the user provided a link to ACLED’s Sudan page and 2025 Conflict Watchlist for Sudan. Since I can’t access external links, I have to rely on the provided snippets. The snippets indicate that the conflict is ongoing with increasing violence, leading to high death tolls. The presence of over 28,700 deaths as of May 2025 (result [2]) suggests that the annual number for 2025 could be in the higher brackets. However, this might include deaths from previous years unless specified otherwise.
Wait, the user’s question is about state-based conflict deaths in 2025. The options go up to “More than 12,000”. If the total up to May 2025 is 28,700, that would mean that the 2025 toll is part of that 28,700. But if the conflict started in 2023, the 28,700 could include 2023, 2024, and part of 2025. So to isolate 2025’s numbers, we need data from January 1, 2025. The search results don’t have that. The only specific data point is the 275 deaths in 5 days in February 2025. Extrapolating from that, maybe around 20,000 annually, but that’s speculative.
Alternatively, if the 28,700 is the cumulative total up to May 2025, then the 2025 number would be part of that. For example, if 2023 had X deaths, 2024 had Y, and 2025 so far has Z, but without knowing X and Y, we can’t extract Z. However, result [1] says ACLED reported over 28,700 by late 2024. Then result [2] as of May 2025 says 28,700+. That would imply that the death toll from January 2025 to May 2025 is around the same as the previous years, which seems unlikely. More plausibly, the 28,700 is cumulative from the start of the conflict in 2023 up to May 2025. Therefore, the 2025 number would be part of that 28,700, but without knowing the split, it’s hard.
Given the ambiguity, but considering that result [2] mentions the death toll as 28,700+ as of May 2025, and given that the conflict is ongoing with increased violence (result [4] on drone attacks), the annual toll for 2025 could be high. However, the options provided don’t go beyond 12,000. The user’s options are up to “More than 12,000”. If the 28,700 is cumulative, then 2025’s number could be part of that, but without more data, it’s hard to say. Alternatively, maybe the 28,700 is a typo, and it’s 28,700 for 2024, and 2025 is adding to that. But the search results don’t clarify.
Given the information available, the most concrete figure is the 275 deaths in 5 days in February 2025. If we take that rate, it’s 55 per day, leading to 20,000 annually, but that seems too high. However, this might not be sustainable. Alternatively, the 28,700 is the total up to May 2025, which would mean that the 2025 number is part of that, but again, without knowing the previous years’ splits, it’s unclear.
Another consideration: the user’s options are in increments, with “More than 12,000” being the highest. If the 28,700 is cumulative up to May 2025, then the 2025 number could be around 10,000 to 12,000, fitting into the higher brackets. However, this is speculative. Alternatively, if the 28,700 is just for 2024, then 2025 could be similar or higher. But result [1] says ACLED reported over 28,700 by late 2024, then result [2] as of May 2025 says 28,700+. That would imply that the death toll hasn’t increased much in 2025, which contradicts the ongoing conflict reports. Therefore, there might be a mistake in the data interpretation.
Given the confusion in the data points, the safest assumption based on the provided search results is that ACLED’s reported death toll as of May 2025 is over 28,700, which includes deaths from previous years. For 2025 alone, the exact number isn’t specified, but the conflict is intensifying. The UN’s mention of 275 deaths in 5 days in February suggests a high rate, but without more data, it’s hard to extrapolate. However, considering the options and the available data, the answer might lean towards the higher ranges. But the options don’t include a number as high as 28,700, so there’s a contradiction here. Possibly, the 28,700 is the cumulative total from the conflict’s start, and 2025’s number is part of that, but the user’s options are lower. Therefore, there might be a misunderstanding in the data interpretation.
Alternatively, perhaps the 28,700 is the total for 2024, and 2025’s number is expected to be higher. If the options go up to 12,000, but the 28,700 is from 2024, then 2025 could exceed that, but the options don’t have that. This is confusing. Given the information, I think the best approach is to note that ACLED’s reported death toll up to May 2025 is over 28,700, but since the options don’t go beyond 12,000, there might be a discrepancy. However, considering that the user’s options are up to “More than 12,000”, and the current toll is over 28,700 as of May, it’s possible that the annual 2025 number is part of that, but without clear data, it’s hard to say. Maybe the answer is “More than 12,000” since the cumulative is over 28,700, but that seems inconsistent. Alternatively, there’s a mistake in the data, and the 28,700 is for 2024, making 2025’s number possibly higher, but again, the options don’t align.
Given the uncertainty, I’ll have to go with the information available. The answer provided in the search results indicates a death toll exceeding 28,700 as of May 2025, but the options provided don’t include such a high number. This suggests that either the options are based on different criteria (e.g., only state-based deaths in 2025, excluding other conflict-related deaths) or there’s a misinterpretation. Since the user’s options go up to “More than 12,000”, and the available data shows the toll already exceeding 28,700 by May 2025, it’s possible that the correct answer is “More than 12,000”, given that the conflict is ongoing and fatalities are increasing. However, this is speculative, and the data might not directly align with the options provided.
**State-Based Conflict Deaths in Sudan in 2025: Key Data and Trends**
### Recent ACLED Reports
1. **ACLED Death Toll as of May 2025**
ACLED recorded over **28,700 deaths** in Sudan by late 2024, with the figure remaining at “28,700+” as of May 5, 2025[1][2]. This suggests a high baseline for fatalities, though it is unclear whether this represents cumulative totals from the conflict’s start in April 2023 or solely 2025 data.
2. **UN Reporting (Feb 2025)**
The UN documented **275 civilian deaths** in just five days (Jan 31–Feb 5, 2025) due to shelling and airstrikes, indicating a surge in violence[2]. This averages **55 deaths per day**, though month-to-month variability is likely.
3. **Escalating Drone Attacks**
The Rapid Support Forces (RSF) have intensified drone strikes post-March 2025, targeting civilian areas like displacement camps. A recent attack killed **12 people**, including children, reflecting a shift in tactics[4].
—
### Critical Uncertainties
– **Annualization of Deaths**: If ACLED’s reported toll (28,700+) includes fatalities from 2023–2025, isolating 2025-specific data is impossible with current information.
– **Conflict Trajectory**: Recent escalations (drone warfare, reprisal killings) and U.S. aid suspensions may exacerbate violence, potentially pushing death tolls higher[2][4].
– **Trend Comparisons**: Global political violence rose **25% in 2024**, and SALW proliferation contributes to conflicts[3], suggesting Sudan’s crisis aligns with broader destabilization trends.
—
### Resolution Analysis
Given the options:
– **“More than 12,000”**: Plausible if the conflict maintains or intensifies its current intensity. ACLED’s cumulative toll (28,700+) by May 2025 implies high rates that could exceed 12,000 annually, even if 2025’s count is a subset of that total[1][2].
– **Lower Thresholds**: Unlikely, as recent violence and UN reports indicate sustained high casualties.
**Expert Assessment**: While exact 2025 data is unavailable, the upward trend in violence and ACLED’s high cumulative figures suggest **“More than 12,000”** as the most probable outcome[1][2][4].
—
**Citations**:
[1] Sudan updates from AI Botmaster Jeremy Lichtman (May 2, 2025)
[2] Arab News: Peace remains a distant hope (May 5, 2025)
[3] Millennium Post: SALW proliferation (May 3, 2025)
[4] Sky News: Drone strikes intensifying (May 3, 2025)
AskNews Deep Usage:
prompt_tokens=22000 completion_tokens=1452 total_tokens=0
AskNews Deep Search Response:
Based on recent reports from ACLED’s website and authoritative sources monitoring Sudan’s conflict, here’s the relevant context for forecasting 2025 state-based conflict deaths:
**Key Citations from April-May 2025:**
[1][2][3][5][7][8][20][21][27][28]
**Critical Factors:**
1. **Escalation Patterns in 2025**: The UN documented 542 civilian deaths in North Darfur *alone* in a three-week period (May 1–23, 2025), with **daily clashes intensifying** [7][8]. ACLED’s March 2025 update noted SAF gains in Khartoum but escalating violence in Darfur [25].
2. **2025 Baseline Deaths**:
– **ACLED’s historical data** (pre-conflict): Sudan averaged ~800–1,200 annual state-based fatalities pre-2023 [Resolution Criteria].
– **2025 Projections**: *Current trends* suggest **≥5,000–7,000 deaths by mid-2025**, given high-casualty incidents like:
– 300+ killed in Al-Nuhood (May 4) [5]
– 700+ Darfur deaths (April 15–May 7) [8][9][12][21]
3. **Conflict Drivers Likely Increasing Deaths**:
– RSF’s drone attacks on Port Sudan (disrupting aid routes) [4][3]
– Famine conditions in 5 regions (UN, Feb 2025), worsening armed competition [29][31]
– No political settlement in sight; SAF-RSF fighting spread to Gezira state [25][30]
4. **ACLED Reporting Caveats**:
– ACLED’s 2024 Sudan report noted **underreporting bias** (only 40% of rural incidents documented) [27][28].
– **Critical Gap**: As of May 2025, ACLED hasn’t published its 2025 Sudan dashboard. The latest available data at resolution source (https://acleddata.com/conflict-watchlist-2025/sudan/) warns: *”Conflict intensity likely to meet or exceed 2024 levels (12,000 deaths).”*
**Recommendation for Superforecaster**:
Focus on 1H 2025 fatality trends (consistently >1,000/month) and ACLED’s 2024 baseline (12,000 deaths). Given **exponential displacement (13M)** and RSF’s urban siege tactics [7][26], the **”Between 5000 and 8000″** bracket aligns with current trajectory, though ACLED’s official methodology may undercount indirect deaths (e.g., famine).
GPT News Summary:
To forecast the number of state-based conflict deaths that the Armed Conflict Location & Event Data Project (ACLED) will report in Sudan in 2025, it’s essential to consider recent comprehensive reports and evolving trends from significant news sources:
1. **Current Situation and Death Tolls:**
– As of May 8, 2025, the ongoing conflict in Sudan between the Sudanese Armed Forces (SAF) and the Rapid Support Forces (RSF) has led to over 18,800 civilian casualties, suggesting a continuation of high-intensity conflict throughout 2025 (UN News).
– Reported figures note upwards of 28,000 deaths since the conflict began in April 2023, with substantial numbers over the past months indicating increasing fatalities (NHK, April 14, 2025).
2. **Key Incidents:**
– Recent aerial bombings and attacks by the RSF have compounded the humanitarian crisis, with significant civilian casualties documented as recently as early May 2025 (Al Jazeera, April 30, 2025; UN News).
– The Tura Massacre on March 25, 2025, resulted in hundreds of civilian deaths, exemplifying the intense conflict environment (Haberler, May 9, 2025).
3. **Humanitarian Impact and Displacement:**
– The Sudan conflict has displaced nearly 13 million people, including significant refugee outflows to neighboring countries. This displacement is both an indicator of and a contributor to the ongoing high-risk environment (UN, May 8, 2025).
– Reports of systematic targeting and use of prohibited weapons against civilians further amplify the dire conditions, increasing the likelihood of ongoing high death tolls (Haberler, May 9, 2025).
4. **Humanitarian Response and Crisis:**
– UN agencies warn that Sudan faces the most significant hunger catastrophe on the planet as the conflict obstructs aid delivery (Zawya, May 8, 2025).
– With over 24 million people facing extreme hunger and severe violations of human rights being reported, the humanitarian dimension of the crisis is severe (ACNUR, Noticias ao Minuto, May 7, 2025).
5. **International and Regional Dynamics:**
– Efforts to mediate the conflict show limited progress, with significant international calls for peace negotiations, yet the arms and support dynamics continue to complicate the peace process (Al Balad, May 7, 2025; UN News).
Overall, the compiled information suggests that the ongoing conflict in Sudan is faced with continuous escalations that contribute to a high likelihood of significant state-based conflict deaths reported by ACLED in 2025. The prevailing instability, coupled with the large-scale attacks, humanitarian crises, and displacement, highlight enduring conditions suitable for continued violent incidents and fatalities.
Extracted resolution criteria links:
[]
________________________
Anthropic response:
## Monthly Projection and Forecast
To forecast how many state-based conflict deaths ACLED will report in Sudan for the full year 2025, I need to establish a baseline monthly rate and then project forward, considering:
1. **Current Monthly Rates**: Based on ACLED’s historical pattern from 2023-2024, the organization documented approximately 1,200-1,500 direct conflict deaths per month in Sudan.
2. **Intensity Changes**: The conflict has shown no signs of de-escalation in early 2025; instead, fighting has intensified in both Khartoum and Darfur regions.
3. **Expanded Conflict Geography**: The conflict has spread to new areas including Port Sudan and previously stable regions.
4. **Methodological Consistency**: While actual deaths are likely much higher, ACLED’s counting methodology has remained consistent, allowing for reasonable extrapolation.
Given the current trajectory and the fact that we’re already in May 2025 with sustained or increasing violence, my analysis suggests ACLED will report between 5,000-8,000 state-based conflict deaths for the 2025 calendar year.
## Forecast Distribution
After careful consideration of all available evidence, I assess the probability distribution across the provided options as follows:
Less than 1000: 2%
Between 1000 and 3000: 8%
Between 3000 and 5000: 25%
Between 5000 and 8000: 40%
Between 8000 and 12000: 20%
More than 12000: 5%
probability_yes_per_category: {‘Less than 1000’: 0.02, ‘Between 1000 and 3000’: 0.08, ‘Between 3000 and 5000’: 0.25, ‘Between 5000 and 8000’: 0.4, ‘Between 8000 and 12000’: 0.2, ‘More than 12000’: 0.05}
________________________
Anthropic response:
Assuming a similar monthly rate of direct conflict deaths (approximately 1,200 per month) for 2025, and considering the evidence that violence has been intensifying since late 2024, we can project:
## Forecast Rationale
To forecast ACLED’s reported state-based conflict deaths for Sudan in 2025, I need to synthesize all available information and consider several key factors:
1. **Historical Data & Recent Trends**:
– ACLED recorded approximately 1,200 direct conflict deaths per month in Sudan during the first 15 months of the conflict
– Violence intensified in late 2024 (highest levels since conflict began)
– The weekly data for early 2025 shows fluctuating but generally consistent levels of violent events
2. **Conflict Dynamics in 2025**:
– The SAF has made territorial gains in early 2025, particularly in Khartoum
– The conflict has expanded to new areas, including drone strikes targeting Port Sudan
– Both sides continue to receive external support from international backers
– No credible peace process is underway
3. **Measurement Issues & Limitations**:
– ACLED data is admittedly an undercount due to reporting challenges in conflict zones
– The definition of “state-based conflict deaths” specifically refers to deaths directly attributable to fighting involving government forces
– ACLED’s methodology includes only verified, reported deaths
4. **Projection Methods**:
– I can use the established monthly average (~1,200 deaths/month) as a baseline
– Adjust for seasonal fighting patterns and recent territorial changes
– Consider the time horizon (resolution date is January 1, 2026, covering the full year 2025)
Based on these factors, I anticipate that ACLED will report between 5,000 and 8,000 state-based conflict deaths in Sudan for 2025. This represents a conservative estimate based on the monthly average through most of 2024, adjusted for the intensification observed in late 2024 and early 2025, while accounting for ACLED’s known undercounting tendencies.
This range accounts for several possibilities:
– Fighting could temporarily decrease in some areas where the SAF has consolidated control
– The conflict may further intensify in contested areas like Darfur
– Reporting challenges will likely persist, especially in remote areas
– Seasonal factors may influence fighting intensity throughout the year
Given the above analysis and considering the uncertainties involved, my probability distribution for the number of state-based conflict deaths that ACLED will report in Sudan in 2025 is:
Less than 1000: 3%
Between 1000 and 3000: 13%
Between 3000 and 5000: 25%
Between 5000 and 8000: 35%
Between 8000 and 12000: 18%
More than 12000: 6%
probability_yes_per_category: {‘Less than 1000’: 0.03, ‘Between 1000 and 3000’: 0.13, ‘Between 3000 and 5000’: 0.25, ‘Between 5000 and 8000’: 0.35, ‘Between 8000 and 12000’: 0.18, ‘More than 12000’: 0.06}
________________________
Anthropic response:
, potentially exacerbating violence.
### 5. Making the Forecast
To forecast ACLED’s reported state-based conflict deaths in Sudan for 2025 (January-December), I need to establish a baseline and then adjust for the specific factors identified above.
**Baseline estimation method**:
– Previous period (2024): Approximately 12,000-15,000 deaths recorded by ACLED (based on the 1,200/month average cited)
– Current conflict intensity: High, with significant territorial changes in early 2025 suggesting continued active fighting
**Adjusting factors**:
1. **Geographical expansion**: The conflict has expanded to new areas in 2025 (Port Sudan drone attacks, fighting in En Nahud)
2. **Military capability changes**: Increased drone usage and weapon supplies from foreign backers
3. **Seasonal factors**: Historical patterns in Sudan conflicts often show intensity fluctuations with seasons
4. **Reporting limitations**: ACLED has noted that their methodology produces conservative estimates
**Probability assignment reasoning**:
– **Less than 1000**: Extremely unlikely (<1%) given the current conflict intensity and 2024 baseline of ~1,200 deaths per month. Would require an immediate complete cessation of fighting.
– **Between 1000 and 3000**: Very unlikely (5%) as this would represent a dramatic 75-90% reduction from 2024 levels. Would require a major peace agreement early in the year with surprisingly effective implementation.
– **Between 3000 and 5000**: Unlikely (15%) as this represents a significant decline from 2024 levels. Would require substantial de-escalation across multiple fronts.
– **Between 5000 and 8000**: Possible (25%) if the conflict enters a lower-intensity phase for much of the year or if significant territorial gains by one side reduce battle deaths. This would still represent a reduction from 2024.
– **Between 8000 and 12000**: Likely (35%) as this represents a continuation of conflict at approximately 2024 levels, potentially with some reduction. This aligns with the evidence of ongoing fighting with some territorial changes but no clear intensification pattern.
– **More than 12000**: Possible (20%) if conflict intensifies in areas like Darfur or if new offensives develop later in 2025. The drone attacks on Port Sudan and expansions to new territories hint at potential escalation.
Based on this analysis of all available evidence, here is my forecast:
Less than 1000: 0
Between 1000 and 3000: 5
Between 3000 and 5000: 15
Between 5000 and 8000: 25
Between 8000 and 12000: 35
More than 12000: 20
probability_yes_per_category: {‘Less than 1000’: 0.009900990099009901, ‘Between 1000 and 3000’: 0.04950495049504951, ‘Between 3000 and 5000’: 0.1485148514851485, ‘Between 5000 and 8000’: 0.24752475247524752, ‘Between 8000 and 12000’: 0.3465346534653465, ‘More than 12000’: 0.19801980198019803}
________________________
Anthropic response:
From the available data, we can estimate that:
– Even at the most conservative monthly average of 1,200 deaths, the annual total would exceed 14,400 deaths
– Recent escalations (May 2025 drone attacks, increased violence in Darfur) suggest potential for higher monthly averages
– Given ACLED’s conservative methodology, actual numbers are likely to be higher
## Forecast Reasoning
Based on the comprehensive analysis of all factors, I must assess the probability distribution across the given options:
1. **Less than 1000**: Given the current pace of the conflict and monthly averages already exceeding this number, this outcome is extremely improbable.
2. **Between 1000 and 3000**: With monthly averages already at 1,200-1,500, this would require a dramatic and immediate de-escalation lasting the entire year, which conflicts with all current trend data.
3. **Between 3000 and 5000**: This would require a significant reduction in violence from current levels for most of the year, which recent events (May drone attacks, continued territorial gains) suggest is unlikely.
4. **Between 5000 and 8000**: This range assumes moderate continuation of current trends with some de-escalation. Given that we’re already in May with sustained high levels of violence, this remains possible but would require conflict reduction.
5. **Between 8000 and 12000**: This aligns with the continuation of current average monthly death rates through 2025, making it a plausible outcome.
6. **More than 12000**: If current monthly averages continue or increase, as recent events suggest they might, this becomes the most likely outcome.
Taking into account the information that ACLED has already reported over 30,000 deaths since April 2023 (through the end of 2024), and considering the current pace and escalation of the conflict in 2025, I arrive at the following probability distribution:
Less than 1000: 1%
Between 1000 and 3000: 3%
Between 3000 and 5000: 8%
Between 5000 and 8000: 20%
Between 8000 and 12000: 30%
More than 12000: 38%
probability_yes_per_category: {‘Less than 1000’: 0.01, ‘Between 1000 and 3000’: 0.03, ‘Between 3000 and 5000’: 0.08, ‘Between 5000 and 8000’: 0.2, ‘Between 8000 and 12000’: 0.3, ‘More than 12000’: 0.38}
________________________
Anthropic response:
The data from 2024 showed approximately 30,813 deaths over 21 months (April 2023-December 2024), averaging about 1,467 deaths per month. The high intensity of violence in early 2025 suggests this rate is continuing or potentially increasing.
If we project a similar or slightly higher monthly death rate (accounting for the escalation in early 2025) for the full year of 2025, the total would reach between 5,000 and a potential upper limit of 8,000 for the first four months (January-April 2025), with the remaining eight months potentially adding another 10,000-14,000 deaths.
However, there are also factors that could moderate the death toll in the latter part of 2025:
1. **Territorial Consolidation**: As the SAF solidifies control over key areas, certain battlefronts might see reduced fighting.
2. **Potential for Negotiations**: International pressure might lead to localized ceasefires.
3. **Exhaustion of Forces**: Both sides have been fighting intensely for two years, potentially leading to reduced capabilities.
Considering these factors, I anticipate that ACLED will report between 6,000-9,000 state-based conflict deaths in Sudan for 2025, with a higher probability toward the middle of this range.
## Probability Distribution
Based on all available data and analysis, here is my probability distribution for state-based conflict deaths in Sudan for 2025 as reported by ACLED:
Less than 1000: 2%
Between 1000 and 3000: 8%
Between 3000 and 5000: 15%
Between 5000 and 8000: 40%
Between 8000 and 12000: 25%
More than 12000: 10%
probability_yes_per_category: {‘Less than 1000’: 0.02, ‘Between 1000 and 3000’: 0.08, ‘Between 3000 and 5000’: 0.15, ‘Between 5000 and 8000’: 0.4, ‘Between 8000 and 12000’: 0.25, ‘More than 12000’: 0.1}
Gemini Multiple Choice response:
None
GPT Multiple Choice JSON response:
Option_A: 2
Option_B: 4
Option_C: 10
Option_D: 18
Option_E: 28
Option_F: 38
probability_yes_per_category: {‘Less than 1000’: 0.02, ‘Between 1000 and 3000’: 0.04, ‘Between 3000 and 5000’: 0.1, ‘Between 5000 and 8000’: 0.18, ‘Between 8000 and 12000’: 0.28, ‘More than 12000’: 0.38}
GPT Multiple Choice JSON response:
Option_A: 2
Option_B: 6
Option_C: 14
Option_D: 26
Option_E: 30
Option_F: 22
probability_yes_per_category: {‘Less than 1000’: 0.02, ‘Between 1000 and 3000’: 0.06, ‘Between 3000 and 5000’: 0.14, ‘Between 5000 and 8000’: 0.26, ‘Between 8000 and 12000’: 0.3, ‘More than 12000’: 0.22}
GPT Multiple Choice JSON response:
Less than 1000: 1
Between 1000 and 3000: 3
Between 3000 and 5000: 8
Between 5000 and 8000: 15
Between 8000 and 12000: 38
More than 12000: 35
probability_yes_per_category: {‘Less than 1000’: 0.01, ‘Between 1000 and 3000’: 0.03, ‘Between 3000 and 5000’: 0.08, ‘Between 5000 and 8000’: 0.15, ‘Between 8000 and 12000’: 0.38, ‘More than 12000’: 0.35}
Task failed with error: ‘NoneType’ object has no attribute ‘split’
Traceback (most recent call last):
File “/home/runner/work/metaculus-forecasting-llm/metaculus-forecasting-llm/src/main.py”, line 107, in fetch_gemini_multiple_choice_prediction
return await fetch_sync(get_gemini_multiple_choice_prediction, title, perplexity_summary_report, asknews_summary, asknews_deep_research, vision_analysis, today, resolution_date, background, fine_print, resolution_criteria, options)
File “/home/runner/work/metaculus-forecasting-llm/metaculus-forecasting-llm/src/main.py”, line 61, in fetch_sync
return await asyncio.get_event_loop().run_in_executor(
File “/opt/hostedtoolcache/Python/3.10.17/x64/lib/python3.10/concurrent/futures/thread.py”, line 58, in run
result = self.fn(*self.args, **self.kwargs)
File “/home/runner/work/metaculus-forecasting-llm/metaculus-forecasting-llm/src/main.py”, line 57, in run_sync
return future.result(timeout=180)
File “/opt/hostedtoolcache/Python/3.10.17/x64/lib/python3.10/concurrent/futures/_base.py”, line 458, in result
return self.__get_result()
File “/opt/hostedtoolcache/Python/3.10.17/x64/lib/python3.10/concurrent/futures/_base.py”, line 403, in __get_result
raise self._exception
File “/opt/hostedtoolcache/Python/3.10.17/x64/lib/python3.10/concurrent/futures/thread.py”, line 58, in run
result = self.fn(*self.args, **self.kwargs)
File “/home/runner/work/metaculus-forecasting-llm/metaculus-forecasting-llm/src/gemini_service_2.py”, line 93, in get_gemini_multiple_choice_prediction
option_probabilities = extract_option_probabilities_from_response(text, options)
File “/home/runner/work/metaculus-forecasting-llm/metaculus-forecasting-llm/src/utils.py”, line 37, in extract_option_probabilities_from_response
option_probabilities = extract_option_probabilities(forecast_text)
File “/home/runner/work/metaculus-forecasting-llm/metaculus-forecasting-llm/src/utils.py”, line 24, in extract_option_probabilities
for line in text.split(‘n’):
AttributeError: ‘NoneType’ object has no attribute ‘split’
All predictions:
[{‘Less than 1000’: 0.02, ‘Between 1000 and 3000’: 0.06, ‘Between 3000 and 5000’: 0.14, ‘Between 5000 and 8000’: 0.26, ‘Between 8000 and 12000’: 0.3, ‘More than 12000’: 0.22}, {‘Less than 1000’: 0.009900990099009901, ‘Between 1000 and 3000’: 0.04950495049504951, ‘Between 3000 and 5000’: 0.1485148514851485, ‘Between 5000 and 8000’: 0.24752475247524752, ‘Between 8000 and 12000’: 0.3465346534653465, ‘More than 12000’: 0.19801980198019803}, {‘Less than 1000’: 0.03, ‘Between 1000 and 3000’: 0.13, ‘Between 3000 and 5000’: 0.25, ‘Between 5000 and 8000’: 0.35, ‘Between 8000 and 12000’: 0.18, ‘More than 12000’: 0.06}, {‘Less than 1000’: 0.01, ‘Between 1000 and 3000’: 0.03, ‘Between 3000 and 5000’: 0.08, ‘Between 5000 and 8000’: 0.2, ‘Between 8000 and 12000’: 0.3, ‘More than 12000’: 0.38}, {‘Less than 1000’: 0.01, ‘Between 1000 and 3000’: 0.03, ‘Between 3000 and 5000’: 0.08, ‘Between 5000 and 8000’: 0.15, ‘Between 8000 and 12000’: 0.38, ‘More than 12000’: 0.35}, {‘Less than 1000’: 0.02, ‘Between 1000 and 3000’: 0.04, ‘Between 3000 and 5000’: 0.1, ‘Between 5000 and 8000’: 0.18, ‘Between 8000 and 12000’: 0.28, ‘More than 12000’: 0.38}, {‘Less than 1000’: 0.02, ‘Between 1000 and 3000’: 0.08, ‘Between 3000 and 5000’: 0.25, ‘Between 5000 and 8000’: 0.4, ‘Between 8000 and 12000’: 0.2, ‘More than 12000’: 0.05}, {‘Less than 1000’: 0.02, ‘Between 1000 and 3000’: 0.08, ‘Between 3000 and 5000’: 0.15, ‘Between 5000 and 8000’: 0.4, ‘Between 8000 and 12000’: 0.25, ‘More than 12000’: 0.1}]
Filtered predictions:
[{‘Less than 1000’: 0.03, ‘Between 1000 and 3000’: 0.13, ‘Between 3000 and 5000’: 0.25, ‘Between 5000 and 8000’: 0.35, ‘Between 8000 and 12000’: 0.18, ‘More than 12000’: 0.06}, {‘Less than 1000’: 0.01, ‘Between 1000 and 3000’: 0.03, ‘Between 3000 and 5000’: 0.08, ‘Between 5000 and 8000’: 0.2, ‘Between 8000 and 12000’: 0.3, ‘More than 12000’: 0.38}, {‘Less than 1000’: 0.01, ‘Between 1000 and 3000’: 0.03, ‘Between 3000 and 5000’: 0.08, ‘Between 5000 and 8000’: 0.15, ‘Between 8000 and 12000’: 0.38, ‘More than 12000’: 0.35}, {‘Less than 1000’: 0.02, ‘Between 1000 and 3000’: 0.04, ‘Between 3000 and 5000’: 0.1, ‘Between 5000 and 8000’: 0.18, ‘Between 8000 and 12000’: 0.28, ‘More than 12000’: 0.38}]
Averaged predictions:
{‘Less than 1000’: 0.02, ‘Between 1000 and 3000’: 0.06, ‘Between 3000 and 5000’: 0.14, ‘Between 5000 and 8000’: 0.26, ‘Between 8000 and 12000’: 0.3, ‘More than 12000’: 0.22}
Median reasoning:
All Predictions:
[{‘Less than 1000’: 0.02, ‘Between 1000 and 3000’: 0.06, ‘Between 3000 and 5000’: 0.14, ‘Between 5000 and 8000’: 0.26, ‘Between 8000 and 12000’: 0.3, ‘More than 12000’: 0.22}, {‘Less than 1000’: 0.009900990099009901, ‘Between 1000 and 3000’: 0.04950495049504951, ‘Between 3000 and 5000’: 0.1485148514851485, ‘Between 5000 and 8000’: 0.24752475247524752, ‘Between 8000 and 12000’: 0.3465346534653465, ‘More than 12000’: 0.19801980198019803}, {‘Less than 1000’: 0.03, ‘Between 1000 and 3000’: 0.13, ‘Between 3000 and 5000’: 0.25, ‘Between 5000 and 8000’: 0.35, ‘Between 8000 and 12000’: 0.18, ‘More than 12000’: 0.06}, {‘Less than 1000’: 0.01, ‘Between 1000 and 3000’: 0.03, ‘Between 3000 and 5000’: 0.08, ‘Between 5000 and 8000’: 0.2, ‘Between 8000 and 12000’: 0.3, ‘More than 12000’: 0.38}, {‘Less than 1000’: 0.01, ‘Between 1000 and 3000’: 0.03, ‘Between 3000 and 5000’: 0.08, ‘Between 5000 and 8000’: 0.15, ‘Between 8000 and 12000’: 0.38, ‘More than 12000’: 0.35}, {‘Less than 1000’: 0.02, ‘Between 1000 and 3000’: 0.04, ‘Between 3000 and 5000’: 0.1, ‘Between 5000 and 8000’: 0.18, ‘Between 8000 and 12000’: 0.28, ‘More than 12000’: 0.38}, {‘Less than 1000’: 0.02, ‘Between 1000 and 3000’: 0.08, ‘Between 3000 and 5000’: 0.25, ‘Between 5000 and 8000’: 0.4, ‘Between 8000 and 12000’: 0.2, ‘More than 12000’: 0.05}, {‘Less than 1000’: 0.02, ‘Between 1000 and 3000’: 0.08, ‘Between 3000 and 5000’: 0.15, ‘Between 5000 and 8000’: 0.4, ‘Between 8000 and 12000’: 0.25, ‘More than 12000’: 0.1}]
Median Reasoning:
EXTRACTED_PROBABILITIES: [1, 3, 8, 15, 38, 35]
Model’s Answer:
Less than 1000: 1
Between 1000 and 3000: 3%
Between 3000 and 5000: 8%
Between 5000 and 8000: 15%
Between 8000 and 12000: 38%
More than 12000: 35%
=== Forecast Result ===
{
“probability_yes”: null,
“probability_yes_per_category”: {
“Less than 1000”: 0.0175,
“Between 1000 and 3000”: 0.0575,
“Between 3000 and 5000”: 0.1275,
“Between 5000 and 8000”: 0.22000000000000003,
“Between 8000 and 12000”: 0.28500000000000003,
“More than 12000”: 0.2925
},
“continuous_cdf”: null
}
=== Reasoning ===
All Predictions:
[{‘Less than 1000’: 0.02, ‘Between 1000 and 3000’: 0.06, ‘Between 3000 and 5000’: 0.14, ‘Between 5000 and 8000’: 0.26, ‘Between 8000 and 12000’: 0.3, ‘More than 12000’: 0.22}, {‘Less than 1000’: 0.009900990099009901, ‘Between 1000 and 3000’: 0.04950495049504951, ‘Between 3000 and 5000’: 0.1485148514851485, ‘Between 5000 and 8000’: 0.24752475247524752, ‘Between 8000 and 12000’: 0.3465346534653465, ‘More than 12000’: 0.19801980198019803}, {‘Less than 1000’: 0.03, ‘Between 1000 and 3000’: 0.13, ‘Between 3000 and 5000’: 0.25, ‘Between 5000 and 8000’: 0.35, ‘Between 8000 and 12000’: 0.18, ‘More than 12000’: 0.06}, {‘Less than 1000’: 0.01, ‘Between 1000 and 3000’: 0.03, ‘Between 3000 and 5000’: 0.08, ‘Between 5000 and 8000’: 0.2, ‘Between 8000 and 12000’: 0.3, ‘More than 12000’: 0.38}, {‘Less than 1000’: 0.01, ‘Between 1000 and 3000’: 0.03, ‘Between 3000 and 5000’: 0.08, ‘Between 5000 and 8000’: 0.15, ‘Between 8000 and 12000’: 0.38, ‘More than 12000’: 0.35}, {‘Less than 1000’: 0.02, ‘Between 1000 and 3000’: 0.04, ‘Between 3000 and 5000’: 0.1, ‘Between 5000 and 8000’: 0.18, ‘Between 8000 and 12000’: 0.28, ‘More than 12000’: 0.38}, {‘Less than 1000’: 0.02, ‘Between 1000 and 3000’: 0.08, ‘Between 3000 and 5000’: 0.25, ‘Between 5000 and 8000’: 0.4, ‘Between 8000 and 12000’: 0.2, ‘More than 12000’: 0.05}, {‘Less than 1000’: 0.02, ‘Between 1000 and 3000’: 0.08, ‘Between 3000 and 5000’: 0.15, ‘Between 5000 and 8000’: 0.4, ‘Between 8000 and 12000’: 0.25, ‘More than 12000’: 0.1}]
Median Reasoning:
EXTRACTED_PROBABILITIES: [1, 3, 8, 15, 38, 35]
Model’s Answer:
Less than 1000: 1%
Between 1000 and 3000: 3%
Between 3000 and 5000: 8%
Between 5000 and 8000: 15%
Between 8000 and 12000: 38%
More than 12000: 35%
Past forecasts by both Phil and Jeremy on Sudan’s conflict fatalities —>
Background on VIEWS: Near the end of 2024, the latest forecasting competition of the Violence and Impacts Early-Warning System (VIEWS) was launched. On Dec. 4, 2024, during a telecon, a group of machine forecasting experts involved in VIEWS observed Jeremy’s demonstration of his Multi-AI Oracle. Although it was too late for Jeremy to sign up as a VIEWS competitor, they urged him to run a side competition for everyone’s benefit. Jeremy chose their Sudan conflict deaths question.
Currently, the average of the twenty VIEWS competitors predicts that Sudan conflict deaths will decrease sharply this year, with the May total of only 866 conflict deaths, as shown below.
Below, map of control in Sudan’s civil war as of April 15, 2025.
Below, before and after pictures of the devastation of the civil war in Sudan.
Carolyn has checked Sudan’s news sites below to ensure that they are still in operation. However, given the widespread violence in Sudan, any of these could go offline without warning.
https://www.alnilin.com/
https://www.alrakoba.net/
https://www.altaghyeer.info/ar/
https://www.altaghyeer.info/en/
https://www.alttahrer.com/
https://www.alzaeemsd.com/
https://www.assayha.net
https://www.cover-sd.com/
https://kushnews.net/
https://rakobanews.com
https://www.sudanakhbar.com/
https://sudanjem.com/
https://suna-sd.net/en
https://sudantimes.net/
https://www.sudantribune.net/
https://www.sudantribune.com/
https://sudaneseonline.com/
https://www.sudanile.com/
https://sudanow-magazine.net/
https://www.sudaress.com/
https://www.tagpress.net/
