Blog

Latest articles and insights about proxies, web scraping, and internet infrastructure

On September 7, 2026, CipherCue measured 44,143 European companies with detected CDN: 89.6% of them are backed by Cloudflare, in the Netherlands — 95.6%. We analyze the numbers and methodology, compare with data from W3Techs, explain why a single bot score on 46 million requests per second changes the rules for working with address pools, and what to do when one provider is a single point of failure for both blocking and outages.

📅 September 9, 2026
Read Article

Bright Data removed mobile proxies from its lineup, demand has shifted to dozens of small brands, and the markup for "mobility" remains — 2–5 times the price of a residential gigabyte. We analyze six checks that show whether you have a cellular address or a residential one disguised as cellular: ip-api flags, ASN through RIPEstat, reverse DNS records, a second independent source, a rotation test on 60 outputs, and geo verification.

📅 September 8, 2026
Read Article

The dark web service Nexus has been receiving scans of documents in real-time from the contractor for verification IDScan.net for over a year: 153 million driver's licenses, 10 million ID cards, 3 million travel documents. The FBI is conducting an investigation. Let's analyze why a leaked scan is more dangerous than a leaked password and how to avoid creating unnecessary copies of a document.

📅 September 7, 2026
Read Article

Checking a thousand residential addresses across six free databases showed: 80.3% are marked by DNSBL, but none made it to the lists of dangerous infrastructure, and the threat_score from Cloudflare is zero everywhere. Let's analyze why the Spamhaus label on a residential IP is a mail policy rather than a threat, and which metrics truly predict the performance of the pool.

📅 September 6, 2026
Read Article

The parser returned nothing, you are changing proxies — but the issue was with the website redesign. Let's analyze how to distinguish between a ban and layout drift in five minutes, how adaptive selectors work in Scrapling (element fingerprinting in SQLite and similarity search in 2.46 ms), and why the reference should be taken from the same geo from which you later collect data.

📅 September 5, 2026
Read Article

On September 1, 2026, CrowdSec released version 1.8: proof-of-work and browser fingerprinting are now available in the self-hosted WAF. Let's analyze why the detection recognized the uselessness of resident IP and clean TLS, how much the PoW task costs for a human and a bot (0.017 seconds with the native solver versus 2 minutes on a phone), and why the commercial PoW from Kasada is fundamentally more dangerous than the open Anubis.

📅 September 4, 2026
Read Article