BestChange: Причиной нашей блокировки могла стать недобросовестная конкуренция

investing.ruPublished on 2025-02-14Last updated on 2025-02-14

Накануне блокировки сервера сайта BestChange.ru 13 февраля подверглись DDoS-атакам, направленным на перегрузку сетевой инфраструктуры, а также вывод из строя устройств и каналов связи. Кроме того, 10 февраля около 30 медиаресурсов опубликовали материалы, которые, по мнению команды BestChange.ru, выглядят как заказные, содержат недостоверную информацию и рекламу альтернативных платформ.

В негативных публикациях BestChange называли платформой «обмана пользователей» и сервисом, «который не может защитить своих клиентов». В качестве первопричин, повлекших за собой «множество жалоб», приводились отсутствие сервисной поддержки, проблемы с модерацией обменников и гарантиями безопасности, размещение не совсем достоверной информации о курсах криптоактивов.

«Характер публикаций указывает на возможную целенаправленную кампанию по дискредитации бренда BestChange. Массовый постинг мог быть нацелен на формирование впечатления о необходимости замены сервиса на альтернативные решения конкурентов. Рекламные площадки продолжают сообщать о запросах на публикацию ложно-негативных материалов о BestChange и попытках размещения рекламы конкурентов», ― говорится в документе, который сервис предоставил изданию ForkLog.

Представители BestChange не собираются участвовать в «репутационной войне» с конкурентами и намерены использовать законные контрмеры.

Ранее Роскомнадзор заблокировал доступ к BestChange.ru во внесудебном порядке по инициативе Банка России.

Читайте оригинальную статью на сайте Bits.media

Strategy Launches 'Digital Credit Capital Framework': Authorizes Sale of $12 Billion in Bitcoin, Ending the 'Never Sell' Script

Strategic, the world’s largest corporate holder of Bitcoin (formerly MicroStrategy), has dramatically shifted its long-standing “never sell Bitcoin” strategy by announcing a new “Digital Credit Capital Framework” on June 29. This plan authorizes the sale of up to $1.25 billion worth of Bitcoin to raise cash, establishes a $2.55 billion USD reserve, increases the dividend rate on its STRG preferred shares to 12%, and authorizes up to $1 billion each for repurchases of its own digital credit securities and Class A common stock. This pivot comes amid severe financial pressure. The company’s STRG preferred shares are trading at a ~24% discount to their $100 face value, making new issuances difficult and stalling its buy-Bitcoin funding flywheel. Its annualized dividend obligation has surged to ~$1.2 billion. Meanwhile, its MSTR stock has plummeted 36% in eight days, erasing its traditional premium over its Bitcoin holdings per share. In recent weeks, Strategic has already shifted focus from accumulating Bitcoin to bolstering cash reserves by selling its own MSTR shares. The new framework formalizes this defensive turn, aiming to ensure liquidity, cover dividends, and support its securities prices through buybacks. However, the move risks triggering a “death spiral” if Bitcoin sales pressure the market, further devaluing the company’s core asset. The company also faces a potential securities investigation and carries significant debt, with Bitcoin’s current price below its average acquisition cost.

marsbit2m ago

Strategy Launches 'Digital Credit Capital Framework': Authorizes Sale of $12 Billion in Bitcoin, Ending the 'Never Sell' Script

marsbit2m ago

Vitalik Buterin Says Obfuscation Could Boost Blockchain Privacy

Ethereum co-founder Vitalik Buterin published an article exploring how cryptographic obfuscation could enhance blockchain privacy and security. Obfuscation protects the code of a program itself, making it unreadable while allowing it to function normally, unlike encryption which secures data. Buterin explained that combining obfuscation with blockchain's inherent ability to record ownership could create systems where user privacy is maintained and reliance on central authorities is reduced, enabling secure payments, finance apps, and confidential operations. However, major technical hurdles remain. While research in "indistinguishability obfuscation" has proven it theoretically possible, current implementations are extremely inefficient, with some methods taking impractically long times to execute. Scientists are working on optimizing cryptographic methods and mathematical tools to improve efficiency. Buterin concluded that practical, widely usable obfuscation will take considerable time to develop, but eventual progress could lead to highly secure blockchain products that operate without trusted third parties.

TheNewsCrypto4m ago

Vitalik Buterin Says Obfuscation Could Boost Blockchain Privacy

TheNewsCrypto4m ago

Deforming the Transformer, LLMs Become Smarter

A new research paper proposes "Tapered Language Models (TLMs)," a method that improves large language model performance without adding any parameters. It challenges the standard Transformer design where each layer has the same number of parameters ("feed-forward network" width). Building on evidence that layers are not equally important—earlier layers handle foundational information like grammar, while later layers often reinforce existing judgments—the researchers suggest reallocating model capacity from later to earlier layers. The core idea is to make the layer width taper off monotonically from start to end, keeping total parameters and compute constant. Experiments compared linear, cosine, and sigmoid tapering curves on a 440M parameter model. The cosine curve (e.g., starting width 1.5x baseline, ending 0.5x) achieved the best result, reducing perplexity by 1.84 points compared to the uniform baseline—a significant gain at zero cost. This finding proved robust across four different model architectures (including gated attention and memory-augmented models) and at larger scales (760M and 1.3B parameters), consistently improving performance on commonsense reasoning and language modeling tasks without harming long-context retrieval ability. The work highlights a long-overlooked design dimension: optimal parameter allocation across depth. It offers a "free lever" for efficiency, potentially applicable beyond language models to vision Transformers and diffusion models. The study was conducted by researchers from Mila, Cornell University, and the University of Montreal.

marsbit12m ago

Deforming the Transformer, LLMs Become Smarter

marsbit12m ago

From SpaceX to Galaxy Digital: A Detailed Look at 37 New AI Companies and 7 Crypto Dark Horses Added to the Russell Indexes

On June 26th, following its annual reconstitution, the Russell US Indexes finalized their new components, with changes taking effect for market trading on June 29th. The Russell 3000 Index, representing approximately 98% of the investable US equity market, saw significant turnover. A record $334 billion was traded during Nasdaq's closing cross on reconstitution day, highlighting the massive passive fund flows tied to these benchmarks. Companies newly added to the index are set to benefit from mandatory buying by these funds. The reconstitution raised the market cap threshold between the large-cap Russell 1000 and small-cap Russell 2000 by 24% to $5.7 billion. Overall, 224 new companies entered the Russell 3000. Of these, 19 joined the Russell 1000, and 205 joined the Russell 2000, while 118 firms were removed. Notably, among the newcomers, approximately 37 are companies operating in the AI and semiconductor ecosystem, accounting for roughly 17% of new additions. The most prominent is SpaceX, which, following its recent IPO and soaring valuation, was fast-tracked directly into the Russell 1000 and Top 200 indexes. Additionally, about 7 cryptocurrency-related companies were newly included, representing about 3% of new entrants. These include Galaxy Digital, Bitmine, and Tron, among others. The inclusion of several Decentralized Autonomous Trust (DAT) entities signals the model's sustained market presence. For these smaller AI and crypto firms, index inclusion boosts visibility, potentially attracting further institutional investment and supporting their stock performance.

Odaily星球日报16m ago

From SpaceX to Galaxy Digital: A Detailed Look at 37 New AI Companies and 7 Crypto Dark Horses Added to the Russell Indexes

Odaily星球日报16m ago

Token Uneconomical

"Token Inefficiency" explores the rising economic burden of AI model token usage in enterprises, where escalating costs often fail to match tangible productivity gains. Major companies like Microsoft, Uber, and Meta are facing "token inefficiency"—characterized by budget overruns for tools like Claude Code with unclear returns. This inefficiency stems from supply-side factors like strategic model price hikes by leaders (e.g., Anthropic) and price increases in budget-friendly models, alongside technical waste in Agent systems through context traps, tokenizer inflation, redundant skill calls, and multi-Agent coordination overhead. A deeper demand-side challenge limits token value: their primary utility remains confined to highly digitalized domains like programming, which benefits from automatic, low-cost feedback loops. Extending tokens to physical world tasks or less digitalized industries faces the "Sim-to-Real Gap," where real-world validation is costly and slow, unlike in code compilation. The article warns that this inefficiency concentrates financial risk in mid-tier model developers, potentially fueling circular financing schemes and shadow credit bubbles. It also highlights societal externalities, as data center expansion strains local power grids and inflates utility costs for residents. To achieve a positive net token economy, the path forward requires dual efforts: technical optimizations (context compression, skill reduction, model routing, budget constraints) and business-side discipline (governance, cost attribution, ROI focus). The ultimate goal is shifting from showcasing AI capabilities to maximizing value per token, finding scalable commercial applications that justify the investment and bridge the digital-physical divide.

marsbit34m ago