GCE O/L Information & Communication Technology common mistakes
GCE O/L Information & Communication Technology · Understand the traps and how to avoid them.
When converting between binary, octal, decimal and hexadecimal, candidates convert only the integer part and either drop the fractional part or convert it in the wrong direction (for example turning 0.625 into .625 as if it were already binary).
School practice concentrates on whole-number conversions, so the multiply-by-2 / divide-by-2 routine for fractions is never automated. Under time pressure candidates apply the integer algorithm to the digits after the point, or assume the fraction is unchanged because it 'looks the same'.
What to do instead
Drill the two-direction rule explicitly: integer part divide repeatedly by the target base and read remainders bottom-up; fractional part multiply repeatedly by the target base and read integer parts top-down, stopping when the fraction reaches zero or the required precision. Practise a fixed set such as 0.375, 0.625, 0.1 and 13.8125 across all four bases, and always verify by converting the answer back to decimal.
Candidates state that 1 KB equals 1000 bytes or 1 MB equals 1000 KB when asked about memory or file size, and mix decimal (kilo, mega) prefixes with binary (kibi, mebi) values in the same calculation.
Everyday usage of 'kilo' means 1000 and storage vendors label drives in decimal units, so the decimal meaning feels correct. Textbooks that use 1024 without stating the convention leave the distinction blurred.
What to do instead
Learn the two conventions as separate facts: binary prefixes 1 KiB = 1024 B, 1 MiB = 1024 KiB; decimal prefixes 1 KB = 1000 B, 1 MB = 1000 KB. In exam questions, look for the wording (KB versus KiB, or 'memory' versus 'storage capacity') and show the conversion line before calculating. Practise converting a 4 GiB RAM figure and a 500 GB disk figure into bytes and compare them.
Asked to choose a data type for a field or variable, candidates pick Integer for a telephone number or a postcode, or pick Real for a count of students, and then cannot explain why the choice fails.
Candidates match data types to whether the content 'looks like numbers' rather than to what operations will be performed. They also overlook leading zeros, which are silently lost by numeric types.
What to do instead
Use a decision routine: will arithmetic ever be done on it? If no, it is text (String/Character), regardless of appearance. If yes, is it whole-number counting or measurement? Integer versus Real. Then check for leading zeros, very large values, and fixed-length codes. Apply the routine to a table of fields (phone number, NIC number, price, quantity, date of birth, gender) and justify each choice in one sentence.
Candidates describe the fetch–decode–execute cycle in the wrong order, omit the program counter or instruction register, or claim the ALU fetches instructions from secondary storage.
The cycle is memorised as a phrase rather than as a sequence of register transfers, and the roles of cache, RAM and secondary storage are conflated because all three 'store things'.
What to do instead
Write the cycle as numbered steps naming the registers: PC holds the address, MAR receives it, the instruction is fetched into the MDR then the CIR, the PC increments, the control unit decodes, the ALU executes, results are written back. Draw the von Neumann block diagram beside it and label which bus carries addresses, data and control signals. Then answer 'what happens if the PC is not incremented?' to test understanding.
In network questions, candidates give the function of a hub, switch and router as interchangeable ('connects computers together') and place a router inside a single LAN segment.
All three devices are drawn as boxes between computers in diagrams, and lessons often cover them in one paragraph, so the layer and the addressing method are not linked to the device.
What to do instead
Anchor each device to its address type and layer: hub repeats bits to all ports (no addressing, Layer 1); switch forwards frames using MAC addresses within one LAN (Layer 2); router forwards packets between networks using IP addresses (Layer 3). For each device write one sentence starting 'It reads the ... address'. Then draw a two-LAN topology and mark exactly where each device belongs.
Candidates write SQL or pseudocode that selects columns correctly but omits the WHERE clause, uses = with a wildcard, or puts the aggregate condition in WHERE instead of HAVING.
Candidates translate the English question word by word and stop when the field names are all used, without re-reading the required filter or grouping. Syntax is copied from a single memorised example.
What to do instead
Adopt a fixed translation order: SELECT fields → FROM table(s) → WHERE row conditions → GROUP BY → HAVING group conditions → ORDER BY. Underline the filter words in the question ('only', 'more than', 'each') before writing. Practise three queries on one table: a simple filter, a COUNT with GROUP BY, and a group filter with HAVING, and check each against sample rows by hand.
When tracing an algorithm, candidates execute a loop the wrong number of times (off-by-one), or update a running total before initialising it, and then report the wrong final output.
Candidates read the algorithm rather than simulate it, so they assume the loop behaves as intended instead of following the stated condition and increment. Trace tables are treated as optional decoration.
What to do instead
Always build a trace table with one column per variable plus the condition and output, and fill one row per iteration, including the final failing test. State the initial value of every variable before the first row. After finishing, count the rows to confirm the loop count matches the condition. Practise with a FOR loop from 1 to 5 and a WHILE loop that runs zero times to see the difference.
Candidates label phishing, pharming, spyware and a Trojan as synonyms, or describe a firewall as software that removes viruses.
The terms are taught as a list of threats in one lesson and are all described as 'bad software or messages', so the distinguishing mechanism (deception versus malware versus network control) is never separated.
What to do instead
Build a three-column table: threat, mechanism, and the control that counters it. For example phishing deceives the user into revealing credentials (countered by awareness and link checking), pharming redirects via DNS or hosts-file manipulation (countered by DNS security and certificates), a Trojan disguises itself inside wanted software (countered by antivirus and download policy), and a firewall filters network traffic by rule and does not scan files. Test yourself by writing the mechanism from the name alone.
Candidates describe the waterfall model as allowing the customer to change requirements at any stage, or describe prototyping as having no planning or documentation at all.
The models are memorised as names and one adjective each ('waterfall is rigid', 'agile is flexible'), without linking each stage to its deliverable or to when the customer is involved.
What to do instead
For each model, write the sequence of stages, the main deliverable of each stage, and when user feedback enters. Waterfall: requirements fixed before design, each stage signed off, changes costly after the fact. Prototyping: a working model is built early, shown to the user, and refined iteratively, with documentation produced alongside. Agile: short iterations with working software each cycle. Then answer a scenario question by matching the model to the risk profile of the project.
Candidates answer 'is this legal or ethical?' by declaring any copying illegal, or by treating copyright, patent and trademark as the same protection, without distinguishing personal use, licensing and plagiarism.
The topic is taught through warnings rather than through the specific rights each instrument grants, so candidates collapse all intellectual property into 'you may not copy'.
What to do instead
Learn one line per instrument: copyright protects the expression of a work automatically from creation; a patent protects an invention and must be applied for; a trademark protects a brand identifier. Then separate three questions: is it licensed (site licence, open licence, freeware), is it attributed (plagiarism), and is it lawful (copyright exception). Apply all three to a scenario such as installing one purchased copy on twenty school machines.