Luiz Felipe Barbosa · 12 May 2026 · 13 min read
Telegram has become an unusual common ground in the contemporary global news system. Because the platform imposes minimal editorial constraints and almost no algorithmic curation, news organizations of radically different ownership types now publish to the same broadcast surface, in the same format (channel posts), often in the same language, and frequently to overlapping international audiences. This convergence of form makes Telegram a useful natural laboratory for one of the oldest questions in international media studies: when the platform is held constant, do structurally different ownership regimes still produce measurably different editorial behaviors?
This paper takes that question to a four-channel sample. On one side is PressTV, the English-language broadcaster of the Islamic Republic of Iran Broadcasting (IRIB), a state-funded public-diplomacy outlet whose Telegram channel (t.me/presstv) functions as a global feed for an audience that does not consume Persian-language IRIB content. On the other side are three privately-owned Israeli news channels with overlapping but distinct editorial profiles: ILTV Israel News 24/7 (t.me/iltvnews), Jewish Breaking News or JBN (t.me/jewishbreakingnewstelegram), and The Times of Israel (t.me/thetimesofisrael2022). All four channels publish primarily in English, all four cover overlapping Iran–Israel–US geopolitics, and all four operate in the same Telegram broadcast format. Their primary structural difference is ownership.
To assess whether ownership translates into measurable behavioral signal, I built a Natural Language Processing pipeline that pulls the most recent 1,200 messages from each channel and processes them through a common analytical stack: cadence and volume aggregation, sentiment and emotion classification, BERTopic-style topic modeling, named-entity extraction with co-mention graphs, TF-IDF temporal shift analysis, PMI-ranked phrase signatures, reply-thread metrics, and a media-versus-text comparison. Holding the per-channel sample at 1,200 messages keeps analytical depth roughly equal across the two sides of the comparison, and using a shared pipeline makes findings directly comparable.
The comparison is institutional rather than partisan. The aim is not to adjudicate between Iranian and Israeli framings of the regional conflict, but to measure whether a state-funded foreign-service broadcaster behaves differently on a shared platform than three privately-owned competitors do. Four theoretical lenses guide the interpretation. Herman and Chomsky's (2002) propaganda model identifies ownership, funding, sourcing, flak, and ideology as filters that produce convergence in elite media even under different formal regimes. Hallin and Mancini's (2004) comparative framework distinguishes media systems along axes of state intervention, political parallelism, and journalistic professionalism, and provides typological vocabulary for contrasting an externally-directed state broadcaster with channels embedded in a commercial-partisan hybrid system. Price's (1994) "market for loyalties" frames international broadcasting as a regulated competition among states and other "cartel" members offering packages of identity in exchange for audience allegiance. Nye's (2004) soft-power theory complements this by reading state-funded international broadcasting as an attraction-based instrument of influence, distinct from coercion. Together, the four frameworks predict that the four channels will share a great deal of surface-level content while diverging in three places that ownership most directly shapes: the rhythm of publishing, the structure of topics and actors, and the strategic deployment of media.
Iran and Israel sit at almost opposite ends of Hallin and Mancini's (2004) comparative typology. Iran's broadcasting sector is consolidated under the Islamic Republic of Iran Broadcasting, a constitutionally-protected state monopoly whose director is appointed directly by the Supreme Leader. Press TV, IRIB's English-language television and online service, was launched in 2007 explicitly as an instrument of international communication aimed at non-Persian-speaking audiences (Press TV, n.d.). External assessments consistently describe Iranian broadcasting as a high-state-intervention system in which professional autonomy is structurally limited and political parallelism is extreme: the state and the official ideology are not competing voices in the field but the field itself (Khiabany, 2010; Reporters Without Borders, 2024a). For comparative purposes, this places Press TV near the polar end of Hallin and Mancini's "state intervention" axis and far from any model in which commercial competition or professional norms exert substantial counter-pressure on output.
Israel's media system is structurally very different. Caspi and Limor (1999) characterize the Israeli press as a commercial-partisan hybrid in which a once-strongly-partisan press has been substantially commercialized, but where political parallelism remains higher than in liberal North Atlantic systems and where national-security considerations exert continuous pressure on professional norms (Peri, 2012). Hallin and Mancini's typology, when extended beyond the Western world, locates Israel closer to the polarized-pluralist model than to liberal or democratic-corporatist ones. Critically for this comparison, all three Israeli channels in the sample are privately owned: The Times of Israel is an independent online-first English-language newspaper founded in 2012 by editor David Horovitz and supported primarily by advertising and donor funding (The Times of Israel, n.d.); ILTV is a privately-owned English-language broadcaster oriented toward an international audience; and Jewish Breaking News is a small private aggregator that operates primarily through Telegram and WhatsApp rather than through a traditional newsroom. None of the three is funded or directed by the Israeli state, and none operates under the kind of unitary editorial command that defines IRIB.
Press TV publishes a high-volume English-language news stream whose Telegram channel mirrors the editorial line of its broadcast and web operations. Across the 1,200-message sample, Press TV averaged 150 posts per day across an 8-day window (April 7–14, 2026), with 63.5% of messages carrying media (Press TV Telegram channel, 2026). Editorially, the channel functions as a foreign-service organ: it covers Iranian domestic politics, Iranian leadership, regional conflicts, and Iran's relations with the United States, Israel, and neighboring states for a global English-speaking readership and diaspora — a mission that maps cleanly onto Price's (1994) "market for loyalties" framing.
ILTV Israel News 24/7 is an English-language Israeli broadcaster whose Telegram feed reflects a more conventional editorial cycle. Across its 1,200-message sample, ILTV covered 104 days at an average of 11.5 messages per day, with a 55.2% media share (ILTV Telegram channel, 2026). Jewish Breaking News functions as a faster wire-style service: 25.5 messages per day across only 47 days, with a 44.9% media share, and a heavy use of self-promotional template language directing readers to WhatsApp groups. The Times of Israel is the slowest and most newspaper-like of the three, publishing 9.8 messages per day across 122 days, with every single post carrying a link-preview card from its parent website — a structural artifact that makes its 100% "media share" qualitatively different from the photo and video media used by the other three channels (The Times of Israel Telegram channel, 2026).
The three Israeli channels therefore do not constitute a single editorial type. They differ from each other in cadence, format, and scope. That internal heterogeneity is itself important for the comparison: any finding that distinguishes Press TV from all three Israeli channels simultaneously cannot easily be attributed to the idiosyncrasy of one private outlet, and any finding that splits the Israeli side internally provides evidence about within-system variation under shared ownership conditions.
The proposal identified three axes where public and private ownership should produce the most divergent behavior on a shared platform: cadence, topic structure, and media strategy. The pipeline outputs corroborate that prediction in each axis, while also exposing a substantial layer of cross-channel convergence that the propaganda model anticipates.
The cadence comparison is the cleanest. Press TV behaves like a rolling crisis stream: 150 messages per day, 18-message peak hours, a continuous workday-to-evening dense band, and 75–94% media intensity in its largest hourly spikes (Press TV Telegram channel, 2026). The three Israeli channels, despite drawing the same per-channel sample, occupy three distinct rhythms. ILTV publishes a steady 11.5 messages per day; JBN compresses 1,200 messages into 47 days at 25.5 per day; and The Times of Israel spreads them across 122 days at 9.8 per day. Press TV's per-day rate is therefore roughly 6× ILTV's, 6× JBN's, and 15× The Times of Israel's.
This is exactly the divergence Hallin and Mancini's (2004) state-intervention axis predicts when a foreign-service state broadcaster is compared with private competitors. Press TV does not have to recover its operating costs from a domestic advertising market or subscriber base; it is funded through IRIB's state appropriation and is structured to project Iranian narratives continuously into a global English-language information environment. The three Israeli channels, by contrast, are constrained by newsroom labor costs, audience attention economics, and editorial conventions that govern volume and pacing. The cadence finding therefore operationalizes one of the propaganda model's first filters — the structural relationship between ownership and operating capacity — at the level of measurable post frequency.
The topic-modeling layer reveals an even sharper divergence. Press TV's largest topic cluster across the sample is Topic 2: "leader / khamenei / islamic" at 98 messages, and the second-largest is Israeli / southern / Lebanon at 58. The remaining major clusters cover talks/Islamabad diplomacy (39), Strait of Hormuz strategic geography (36), IRGC posture (34), and occupied Palestine framing (34) (Press TV Telegram channel, 2026). The single most prominent named cluster on Press TV is therefore not war reporting but leadership memorialization — coverage of the 40th day of mourning for Ayatollah Khamenei. This is a structural fingerprint that Price's (1994) market-for-loyalties model predicts directly: a state-funded foreign-service broadcaster would be expected to allocate disproportionate space to symbolic state legitimacy content alongside its conflict reporting, because such content offers the diaspora and sympathetic global audiences a coherent package of identity and allegiance that purely private competitors have no comparable institutional reason to produce.
The three Israeli channels show three different topic structures, none of which centers on memorialization of Israeli leadership. ILTV collapses into a single dominant cluster covering "israel, iran, iranian" at 876 of 910 scored messages — a tight monolithic narrative spine without the symbolic-state-content layer Press TV exhibits. JBN fragments into 19 clusters, the largest of which are dominated by self-promotional WhatsApp/links boilerplate; once that template language is mentally subtracted, the editorial themes are operational reporting on the IDF, Hezbollah, Lebanon, ballistic missile activity, and flight incidents. The Times of Israel fragments into dozens of substantively coherent newsroom clusters: missile-strike reporting, war coverage, West Bank policing, Bondi Beach attack aftermath, Trump–Iran diplomacy, the daily-briefing format, Netanyahu coverage, Gaza/Hamas, Hormuz, Khamenei coverage, and even a research-and-science cluster.
The contrast with Press TV is institutional rather than ideological. Hallin and Mancini's typology helps explain why ILTV's monolithic shape and The Times of Israel's byline-indexed newsroom shape are both visibly Israeli-private products: ILTV operates a smaller, more thematically constrained editorial team focused on a single conflict frame, while The Times of Israel reproduces the section structure and named-byline economy of a professionalized commercial newspaper, with PMI-ranked phrase signatures dominated by recurring journalist names (Ariela Karmel, Rossella Tercatin, Sam Sokol, Amanda Borschel-Dan, Jordan Hoffman). The Times of Israel, in other words, is the channel whose topic model most clearly reflects the Western-style commercial newsroom convention of bylines and beats; Press TV is the channel whose topic model most clearly reflects the convention of state-broadcaster symbolic legitimacy work. Neither convention is visible in the other.
The media-vs-text comparison reinforces the same structural reading. On Press TV, the media/text-only split is editorially meaningful in topic and emotion mix but not in coarse sentiment polarity (p = 0.157 for sentiment score; p < 0.001 for topic). Press TV's media posts over-index strongly on talks/Islamabad diplomacy (+7.7 pp) and the Strait of Hormuz cluster (+5.5 pp), while text-only posts over-index on Khamenei/leader content (-11.3 pp from media's perspective) and Gaza coverage (Press TV Telegram channel, 2026). Visuals on Press TV are, in other words, deployed disproportionately to diplomacy, official actors, and strategic geography — exactly the categories Nye (2004) identifies as soft-power-attractive: official meetings, leaders, and emblematic locations. Symbolic ideological content runs more often as text-only.
The three Israeli channels show three different media patterns, none of which mirrors Press TV's soft-power-style media-for-diplomacy specialization. ILTV's media and text-only posts are editorially almost identical (topic split p = 0.888); the channel uses media mostly to thicken prime-hour coverage rather than to segregate subject matter. JBN shows the largest topic-effect size in the sample (0.44), but the split runs along a private-channel template axis: text-only posts carry the WhatsApp/donate boilerplate, and media posts carry the strike/Hezbollah tactical content. The Times of Israel offers no media/text split at all: every post is a link-preview card to a parent-website article, so its "media share" is structurally fixed at 100%. Across the three Israeli channels, the meaningful axes of variation are operational and commercial; none of them shows the diplomacy-and-leader-visual specialization that Press TV's data exhibits.
Alongside these divergences, the data also shows substantial cross-channel convergence, and that convergence is itself a major comparative finding. All four channels share a fear-dominant emotional baseline. Mean sentiment scores cluster in a narrow band: -0.25 (Press TV), -0.261 (ILTV), -0.227 (JBN), and -0.229 (The Times of Israel). The positive-sentiment label share never exceeds 6.2% on any channel. Fear is the dominant emotion on Press TV (47.2%), ILTV (31.8%), and The Times of Israel (34.6%), and the co-dominant emotion on JBN (36.3%, narrowly behind neutral at 37.5%) (Press TV Telegram channel, 2026; ILTV Telegram channel, 2026; Jewish Breaking News Telegram channel, 2026; The Times of Israel Telegram channel, 2026).
The named-entity layer converges as sharply as the emotion layer does. Each of the four channels produces a top-ten entity list anchored on the same triangle: Iran, Israel, the United States, and their associated demonyms, with Trump, IDF, Hezbollah, Lebanon, and Tehran as stable secondary nodes. The strongest co-mention pair on Press TV is Iran ↔ US (269 co-mentions); the same pair appears in the top tier of every Israeli channel as well. Israel outranks Iran in mentions only on ILTV; Iran outranks Israel on JBN, The Times of Israel, and Press TV. None of the four channels treats the conflict as a domestic story or a regional story disconnected from Washington. All four channels narrate it as a triangular Iran–Israel–US confrontation.
This is precisely the convergence Herman and Chomsky's (2002) propaganda model predicts. Different ownership structures filter content through different mechanisms — direct state direction in Press TV's case, market and political-parallelism pressures in the Israeli channels' case, and shared elite-source dependencies on government and military spokespeople in all four — but the filters terminate in similar topical-affective output: a fear-dominant emotional climate centered on a stable Iran–Israel–US actor cast. The convergence is not evidence that ownership does not matter; the divergences in cadence, topic structure, and media strategy all show that it does. The convergence is evidence that ownership operates within a shared international news environment whose dominant elite sources, framing logics, and event agenda exert pressure on all four channels at once.
Reading the four channels through the four theoretical lenses produces a consistent picture. Hallin and Mancini's (2004) comparative typology is the framework that the data fits most cleanly: the cadence, topic, and media-strategy divergences map directly onto the difference between a high-state-intervention foreign-service broadcaster and three commercial-partisan-hybrid private channels operating under different scales and conventions. Price's (1994) market-for-loyalties argument helps explain why Press TV alone produces a dominant Khamenei/leader memorialization cluster: a state-funded English-language broadcaster competes for global audience allegiance partly by offering legible packages of identity, and ceremonial symbolic content is one such package. Nye's (2004) soft-power framework helps interpret why Press TV's media-vs-text split concentrates visuals on diplomacy, official actors, and strategic geography rather than on ideology. And Herman and Chomsky's (2002) propaganda model explains the strong residual convergence: even when ownership clearly shapes cadence, topic shape, and media use, all four channels still cluster around a fear-dominant emotional baseline and the same Iran–Israel–US actor triangle, because the elite-source environment and the salient event agenda are largely shared.
For globalization theory, the most useful contribution of this comparison is methodological. Telegram is increasingly a place where state-funded international broadcasters and privately-owned commercial outlets meet on equal technical footing, and a common NLP pipeline can measure ownership effects under such conditions in concrete behavioral metrics — messages per day, dominant topic clusters, named-entity co-mention graphs, media-vs-text topic effect sizes — rather than in interpretive claims alone. This does not replace traditional close reading, but it usefully constrains it.
Three reflections close the paper. First, the corpus is small relative to the editorial output of any of these channels in a year, and the temporal window is uneven across channels, with Press TV concentrated in 8 days while The Times of Israel spans 122. Future work should hold the calendar window constant rather than the message count constant, which would convert cadence into a comparison of equal-time output rather than equal-volume output. Second, the analysis here treats Telegram as the unit of observation, but the four organizations also operate websites, broadcast feeds, and other social-media channels. A multi-platform extension would help distinguish Telegram-specific editorial behavior from general organizational behavior. Third, replicating this design for a different conflict cluster — for example, comparing CGTN with privately-owned anglophone Asian outlets, or comparing RT with privately-owned Eastern European outlets — would test whether the convergence-with-divergence pattern observed here generalizes beyond the Iran–Israel–US case.
What the four-channel sample shows most clearly is that ownership does not determine everything but determines important things. On a shared platform, in a shared language, covering a shared event horizon, a state-funded foreign-service broadcaster posts more often, allocates more space to leadership and symbolic content, and uses visuals more strategically for diplomacy than three privately-owned competitors do — even when all four end up emotionally and demographically pointed in the same direction.
A study of 7,320 Kalshi political speech mention contracts showing that these markets learn over time but still overprice low-probability YES outcomes, leaving the NO side materially stronger after fees.
A cross-category study of 18,948 Kalshi mention contracts showing that calibration improves as events approach, but longshot bias still leaves the NO side systematically underpriced.
This project analyzes 84,816 YouTube comments on the 15 most-streamed songs of 2025 to test Adorno’s Culture Industry. It finds that only 0.32% of comments critically engage with the production system behind pop music.