Ranking Algorithms History¶

Timeline of major search engine algorithm updates for both Yandex and Google, including the evolution from heuristic ranking to neural network-based systems.

Yandex Algorithm Timeline¶

Algorithm	Year	Significance
Nepot filter	2005	Nullified link weight when link deemed commercial/paid
Magadan	2008	Named algorithm era begins; geo-dependence; text uniqueness recognition
ACS filter	2009	Sanctions for low-quality auto-generated content
Snozhinks	2009	Matrixnet ML introduced; keyword-stuffed texts penalized
Krasnodor	2010	"Spectrum" technology: classifies queries by different intents
Baden-Baden	2017	Penalty for over-optimized content with excess keyword density
YATI	2020	Transformer-based text analysis; major NLP improvement
Y1	2021	Major update (+2000 improvements); transformer-based YaTI + YaLM

Yandex Neural Network Evolution¶

Pre-2016: Heuristic Era¶

Heuristic algorithms, table-based factors, content-only signals.

2016-2018: Palekh and Korolev¶

Based on Deep Structured Semantic Model (DSSM)
Palekh: compares query against document title
Korolev: extends to full page content comparison; assessor opinions integrated

2019: Expert-Assessment Neural Networks¶

Training targets based on expert quality ratings.

2020+: YATI (Yandex Attention to Intent)¶

Transformer architecture
Multiple streams: anchor list, click-based URL index
Larger text consideration: texts up to 10 sentences processed in full
Word vector space for semantic understanding

Yandex Proxima (Quality Metric)¶

Combines commercial, conversion, expert signals: - Commercial quality components - Domain-specific quality (medicine, legal, financial) - User value signals: Expertise, Authority, Trustworthiness

Google Algorithm Timeline¶

Algorithm	Year	Significance
Panda	2011	Penalizes thin/duplicate/low-quality content; affected 12% of results
Penguin	2012	Penalizes spam links, link farms, purchased link schemes
Hummingbird	2012	Semantic query understanding; context > individual keywords
RankBrain	2015	ML component; relevant pages by meaning without exact match
Mobile-Friendly	2015	Mobile-optimized pages get priority in mobile search
Mobile-First Index	2018	Mobile version used as primary for indexing
BERT	2019	Bidirectional context analysis; prepositions and full sentence meaning
Core Web Vitals	2020	LCP, FID/INP, CLS as ranking signals

Google E-E-A-T Framework¶

Originally E-A-T, expanded to E-E-A-T: - E (Experience) - direct first-hand experience with topic - E (Expertise) - expert-level knowledge of subject - A (Authoritativeness) - authority of author and site in the niche - T (Trustworthiness) - site reliability: quality content, full contacts, payment/delivery info

YMYL sites (health, finance, legal, safety) subject to strictest E-E-A-T requirements since 2018.

Google BERT (2019)¶

Bidirectional Encoder Representations from Transformers: processes query context including prepositions and full sentence, not just individual keywords. Marks shift from keyword matching to semantic understanding.

Key Facts¶

Yandex uses 2000+ ranking factors
Google's algorithm updates have increasingly targeted content quality and user intent
Both engines have converged on transformer-based architectures for semantic understanding
YMYL niches face strictest quality requirements on both platforms

Gotchas¶

Algorithm names matter for diagnostics - when traffic drops, correlating timing with known algorithm rollouts is the fastest way to identify cause
YATI changed everything for Yandex - pre-2020 Yandex text optimization advice is largely outdated
E-E-A-T is not a direct ranking factor - it is a quality framework used by human assessors whose ratings train the algorithms
Mobile-First means mobile IS the index - desktop-only content not visible on mobile version may not be indexed at all