Disclaimer. Since this question is three questions in one (1. what is tadlīs and its types, 2. what is the ruling of a hadith with tadlīs, and 3. what is the ruling of a person who practiced tadlīs), the answer below will scratch the surface of each topic. If you need to get more details about a specific topic, please post a separate question.
Linguistic definition
Tadlīs (Arabic: تدليس) is a word that stems from da-la-sa (Arabic: دلس), which means darkness, according to Lisān al-'Arab 7/86. The word is often translated as fraud (e.g., when a seller intentionally misleads a buyer in a transaction), or as deception (e.g., in cases of camouflage or disguise, etc.).
Hadith terminology definition
From the perspective of the taxonomy of hadith, the word tadlīs is used to refer to a hadith that has a "problem" in its chain of narration, which the narrator hides (thus putting the listener in darkness). Some scholars classify tadlīs into multiple categories, but the most common classification is:
- Tadlīs al-isnād (Arabic: تدليس الإسناد)
- Tadlīs ash-shyūkh (Arabic: تدليس الشيوخ)
A narrator who practices tadlīs is referred to as mudallis (Arabic: مدلس).
Tadlīs is a type of hidden discontinuity (Arabic: السقط الخفي), which makes it difficult for non-scholars of hadith to detect, unlike apparent discontinuity (Arabic: السقط الظاهر), of which its four types (mu'allaq, mursal, munqaṭi', and mu'ḍal) may be detected by non-scholars of hadith. It can be summarized by:
تَدْلِيسُ الاِسْنَادِ بِأَنْ يَرْوِيَ عَنْ ... مُعَاصِرٍ مَا لَمْ يُحَدِّثْهُ بِـ أَنْ
يَأْتِي بِلَفْظٍ يُوهِمُ اتِّصَالا ... كَـ عَنْ وَ أَنَّ وكذاك قالا
وَقِيلَ: أَنْ يَرْوِيَ مَالَمْ يَسْمَعِ ... بِهِ وَلَوْ تَعَاصُرًا لَمْ يَجْمَعِ
Tadlīs al-isnād
Tadlīs al-isnād (Arabic: تدليس الإسناد) may be further classified into four subcategories:
- Tadlīs al-Qat' (Arabic: تدليس القطع)
- Tadlīs as-Sukūt' (Arabic: تدليس السكوت)
- Tadlīs al-'Atf (Arabic: تدليس العطف)
- Tadlīs at-Taswiyah (Arabic: تدليس التسوية).
Tadlīs al-Qat'
Tadlīs al-Qat' is when a narrator narrates a hadith attributed to one of his teachers, but the narrator had not heard that particular hadith directly from his teacher. It is termed al-qat' (severance) as the narrator severed the first indicative words of what took place — for example, "it was narrated to us by", "we were told", "we were informed", "narrated on the authority of", "it was relayed to me", etc., which are often omitted in translations save for the layer of the companions — by saying directly: "Az-Zuhri through Anas said ...".
An example of tadlīs al-qat' is what 'Ali ibn Khashram said that Sufyān ibn 'Uyaynah used to do with his students (among which were Imam Ash-Shāfi'i and Imam Ibn Hanbal) either when testing them or when discussing matters of jurisprudence.
In a long story relayed by Ibn Khashram about Ibn 'Uyaynah testing Ash-Shāfi'i about the lessons learned and deductions that can be made from the hadith in Ṣahih al-Bukhāri 59/90 and Ṣahih Muslim 39/33, he started the narration with "Az-Zuhri through Anas said ...". Az-Zuhri was one of the teachers of Ibn 'Uyaynah. Ibn Khashram asked him three times if he had heard it from Az-Zuhri, after which Ibn 'Uyaynah said that had neither heard it from Az-Zuhri, nor from the one who heard it from Az-Zuhri; rather it was through 'Abdur-Razzāq through Ma'mar through Az-Zuhri through Anas ibn Mālik (Arabic: لا والله ما سمعته من الزهري ولا ممن سمعه من الزهري، بل سمعته من عبد الرزاق عن معمر عن الزهري عن أنس بن مالك رضي الله عنه وأرضاه). This is what is termed the "narration of the equals" (Arabic: رواية الأقران) since both 'Abdur-Razzāq and Ibn 'Uyaynah are from the same layer.
Tadlīs as-Sukūt
Tadlīs as-Sukūt is the one you bring up in your question where the narrator skips the name of the one that had narrated the hadith directly to him, but uses the proper terminology of the action ("it was narrated to us by", "we were told", "we were informed", "narrated on the authority of", "it was relayed to me", etc.). It is termed as-sukūt (silence) as the narrator would say something like: "It was relayed to me by [a pause and a deep breath] Hishām ibn 'Urwa ...". During this silence, and the deep breath, the narrator would say in a very quiet voice, almost always inaudible, the actual name of the person who narrated the hadith to him. The issue here is that the chain of narration seems continuous and proper to the listener. This is what 'Umar ibn 'Ubayd Allah aṭ-Ṭanāfsi did when relaying one hadith, where he said: "It was relayed to me by ... Hishām ibn 'Urwa through his father through 'Ā'isha ...". When asked if he had heard it directly from Ibn 'Urwa, he said no.
Tadlīs al-'Atf
Tadlīs al-'Atf is where the narrator connects two narrators into one — using the word "and" — without skipping any name from the narration chain and while using the proper action terminology ("it was narrated to us by", "we were told", "we were informed", "narrated on the authority of", "it was relayed to me", etc.), whereas he had heard it from only one of them, not both. It is termed al-'atf (connection) since the narrator connects the two narrators using the word "and" rather than the word "through".
For example, this is what Hushaym ibn Bashīr (trustworthy) did with his students one time when they tried to outsmart him. Every time that he started a hadith with "narrated through" (Arabic: عن) without using the proper action terminology, they would ask him if he had truly heard it from that person. So he told them, to teach them how to critique hadith, "It was narrated to me through Huṣayn, and Huṣayn ibn Abdur-Raḥmān is a trustworthy memorizer, narrated to me through Huṣayn and Al-Mughīra, and Al-Mughīra ibn Qāssim is a scholar from the scholars of Kufa, ..." and he continued with the chain of narration. He then asked them if he had practiced any form of tadlīs. They said no, so he then informed them that had not heard from Al-Mughīra, i.e., he heard from Huṣayn who heard from Al-Mughīra.
Tadlīs at-Taswiyah
Tadlīs at-taswiyah is most likely the worst type of tadlīs as the narrator mentions a hadith through one of his teachers using the proper action terminology, but omits mentioning a weak narrator in between two trustworthy narrators from the chain of narration and replaces him with 'an'ana (through-ization, for a lack of a better term, Arabic: عنعنة), i.e., he would drop the proper action terminology in between the two narrators that are incorrectly linked and replaces it with the word "through" or "he said". It is termed at-taswiyah (tidying up) since the chain is apparently "cleaned" of a weak narrator to make it long stronger than it should be. This is typically termed a hadith with a weak chain (Arabic: حديث إسناده ضعيف) by most scholars. Some scholars used a different term to describe such a hadith, e.g., Al-Haythami called it a hadith through trustworthy men in its chain (Arabic: حديث إسناد رجاله ثقات), while Al-Bukhāri termed it a hadith with a non-straight chain (Arabic: حديث إسناده غير مستقيم).
Worth noting is that a hadith with a weak chain (Arabic: حديث إسناده ضعيف) is not necessarily a daʻīf (weak) hadith (Arabic: حديث ضعيف). It becomes a daʻīf hadith only if no other hadiths or other narration chains strengthen it, i.e., it has no other attestations (Arabic: شواهد) or continuations (Arabic: متابعات).
Rulings of scholars on tadlīs al-isnād
Scholars of hadith classified hadiths where tadlīs was practiced into five categories from a biographical evaluation point of view:
- Absolute rejection: Scholars (e.g., Shu'ba, Abdur-Razzāq aṣ-Ṣan'āni, Abu 'Āssim, and Ibn Hazm) who considered any person who practiced tadlīs as majrūḥ (discredited, Arabic: مجروح).
- Absolute acceptance: Scholars (most of the Hanafi school and some of the Maliki school) say that tadlīs is not lying and such a hadith is not necessarily daʻīf; they accept the hadith with a mudallis (Arabic: مدلس) in the narration chain based on a part of an argument made by Ash-Shāfi'i.
- Distinction by narrator and narration: Ash-Shāfi'i said that a tadlīs is not considered a lie; rather, a form of confusion (Arabic: ضرب من الإيهام) that is unacceptable until the narrator explicitly drops al-'an'ana and mentions the narration chain in its entirety. This is conditional upon the narrator being trustworthy in the first place, and that the narration itself does not have 'an'ana. In other words, the person practicing tadlīs is not rejected at large, but his narrations which have tadlīs are rejected. Of course, this is the view most Shāfi'i scholars subscribe to.
- Distinction by the narrator, not by the narration: All narrations of such a narrator (the only one that falls under this category is Sufyān ibn 'Uyaynah) are accepted since the name of the person dropped from a narration chain is also a trustworthy person. In other words, the chain of narration does not get discredited when it is mentioned in its entirety.
- Distinction by the history of the narrator: Scholars accept the hadiths of some narrators who practiced tadlīs on occasions, e.g., narrated a short version of a chain (e.g., to save time) on one occasion but mentioned the full chain on other occasions. This category includes Sufyān ibn 'Uyaynah, Sufyān ath-Thawry, and Az-Zuhri.
Note that some trustworthy narrators practiced 'an'ana through teachers that they spent a long time with, which generally makes this 'an'ana was acceptable. Al-Bukhāri and Muslim, however, rejected such narrations in their Ṣahihs but may have accepted it in other books. This means that they may have included some narrations by one narrator, but excluded some of his other narrations. Adh-Dahabi, in his earlier years, said that a long stint with a teacher by a narrator is a cause for overlooking al-'an'ana, e.g., that of Al-A'mash through Abu Ṣāliḥ (which dropped Abu Wā'il in between). He, later on, retracted this in his book Mīzān al-I'tidāl. In Al-Mustadrak 'ala aṣ-Ṣaḥiḥain, Al-Hākim an-Nisapūri included all hadiths by narrators in Ṣahih al-Bukhāri and Ṣahih Muslim without looking at the context of al-'an'ana, therefore it is not necessarily considered a book of authentic hadiths in its entirety.
Tadlīs ash-shyūkh
Tadlīs ash-shyūkh (Arabic: تدليس الشيوخ) is when a narrator mentions a hadith through his teacher, while using the proper termnology of the action ("it was narrated to us by", "we were told", "we were informed", "narrated on the authority of", "it was relayed to me", etc.), but gives his teacher a name or a nickname that he is not commonly known by, e.g., calling Sufyān ibn 'Uyaynah (the name by which he is commonly known) by Abu Muhammad al-Hilāli (a name by which he is not commonly known). Tadlīs ash-shyūkh is sometimes done as a form of testing the knowledge of a seeker of knowledge. Adh-Dahabi, who was a student of Ibn Taymiyyah and Ibn Daqīq al-'Eid, wrote about an incident when Ibn Daqīq al-'Eid tested him about the name of Sufyān ibn 'Uyaynah in such a fashion.
Another form of tadlīs ash-shyūkh is tadlīs al-bildān (Arabic: تدليس البلدان) in which the narrator narrates through someone who is unknown but shares the same name as a famous person and does so in a place that the narrator implies it to mean he had traveled to the place of the known scholar. For example, a narrator meets someone called Al-Hassan who is from Basra, so he calls him Al-Hassan al-Basri. He then takes him to a street called Basra street (not the actual town of Basra), and say "when I was in Basra, Al-Hassan al-Basri relayed to me ...".
Rulings of scholars on tadlīs ash-shyūkh
If the purpose of tadlīs ash-shyūkh is by a teacher to test his students, then it is acceptable and does not discredit the teacher. If it is to give the impression that the hadith is authentic, then that person is discredited for this hadith as well as other hadiths, e.g., 'Aṭiyyah al-'Oufi who used to drop the name of Muhammad ibn as-Sā'ib and replace it by Abu Sa'id al-Khudri to give credibility to his hadiths.
NOTE. I briefly explained why the hadiths of Sufyān ibn 'Uyaynah are accepted. When it comes to the rulings on specific narrators, this may be the topic of another set of questions, with each specific narrator was accused of tadlīs at one time (e.g., Abu Sa'īd al-Anṣāri, Al-A'mash, Al-Bukhāri, Al-Hajjāj ibn Arta'a, Al-Walīd ibn Muslim, Baqiyyah ibn al-Walīd, Hishām ibn 'Urwah, Ibn Hazm, Ibn Juraij, Muslim, Mālik, Sufyān ath-Thawri, Sufyān ibn 'Uyaynah, Yazīd ibn Harūn as-Wāsiṭi, among others) in a separate question. Being accused of tadlīs means nothing unless the accuser(s) is a reliable and knowledgeable person, and very few among those was actually discredited as a mudallis. Today, we have every Tom, Dick, and Harry having opinions about scholars like Al-Bukhāri or Muslim when they cannot in the first place list their methodology of authentication.